Moral Reasoning

While there has been much discussion of whether AI systems could function as moral agents or acquire sentience, there has been very little discussion of whether AI systems could have free will. I sketch a framework for thinking about this question, inspired by Daniel Dennett's work. I argue that, to determine whether an AI system has free will, we should not look for some mysterious property, expect its underlying algorithms to be indeterministic, or ask whether the system is unpredictable. Rather, we should simply ask whether we have good explanatory reasons to view the system as an intentional agent, with the capacity for choice between alternative possibilities and control over the resulting actions. If the answer is "yes", then the system counts as having free will in a pragmatic and diagnostically useful sense.

Alignment applied after pretraining is shallow in a measurable way: a single direction in a model's residual stream can be edited out, and the model stops refusing harmful requests. That fact says how easily refusal can be removed, not what the refusal decision was reading in the first place. We ask what it reads, and we separate that from what the model comprehends. Across four open-weight models spanning three families, moral comprehension is native to pretraining: a low-rank moral subspace crystallizes during pretraining, and alignment rotates it once without rebuilding it. The refusal gate, in contrast, is a fresh post-training construction with only a weak pretraining precursor, written into a narrow control-token channel where the refusal decision is orthogonal to the moral-judgment decision. The central result is causal and comes from one model, OLMo-3. A nested interchange rank sweep patches successively larger slices of the moral subspace between matched requests and reads how much of refusal's response transfers: as the basis widens, moral judgment keeps reading more of it, while refusal levels off at the level of a single harm direction, and about three-quarters of refusal's causal input lies outside the moral subspace altogether. Refusal reads the harm percept, not the moral content that judgment reads on the same patches. The picture is not uniform across families. Llama reads broad moral content; Qwen reads beyond the single harm cue but is unresolved at our sample size; GPT-OSS reads harm, and its refusals can be argued in either direction by its own reasoning trace. Where refusal reads only a low-rank slice and routes around the bulk of what the model knows, a rank-one edit removes it. Whether widening what refusal reads would also deepen the behavior is the open question this raises.

In recent years, artificial intelligence has made extraordinary progress thanks to large-scale models capable of generalization and the generation of complex outputs. However, transferring this potential into embodied agents reveals a significant limitation: the most advanced systems rely on pre-existing datasets and human feedback strategies that are powerful but insufficient in dynamic or unknown contexts. To adapt, an agent must acquire knowledge through direct interaction with its environment. One strategy to address this challenge involves introducing higher-level mechanisms, such as intrinsic motivations, which leverage curiosity and competence, to guide exploration and learning in complex environments. While this flexibility expands autonomy, it complicates the task of ensuring agents remain aligned with human goals. Alignment, already a challenge for artificial systems in general, becomes even more complex in unstructured and dynamic contexts where predefined rules prove insufficient. To be effective and adaptable, norms must be rooted in experience through an epistemological process that starting from simple, situated principles allows for the gradual construction of more complex rules through experience, autonomous learning, and cooperation with other moral agents. Similarly to children learning social norms by exploring their environment and participating in collective practices, artificial agents must also be educated toward alignment. Following Dennett, the status of a moral agent is not innate but is attributed gradually based on the ability to responsibly manage increasing degrees of freedom. From this perspective, the regulatory sandboxes can be viewed as pedagogical environments for AI: dynamic spaces where alignment develops as a formative process, progressively shaping autonomous behaviors through interaction and cooperation in scenarios of increasing complexity.

Claims that Artificial Intelligence systems improve decisions, broaden access, reduce harm, or empower users can exceed what their evaluation establishes. Predictive performance alone does not establish safety, the presence of oversight does not establish meaningful control, and faster task completion does not establish understanding or choice. Evaluation must account for unreliable outputs and uneven performance, but also for overreliance, weakened recourse, and displaced human expertise. The harder questions are what the evidence warrants, which relations of power remain unexamined, and where measurement must stop. Assessing improvement requires examining what institutions value and the conditions AI is asked to address. AI is both revelation and intervention. Its use can reveal unmet human needs and assumptions about what matters. Once deployed, it can repair, compound, substitute for, or conceal existing failures. We develop a rupture test that evaluates deployment against explicit human and non-AI baselines. Drawing on Pope Leo XIV's Magnifica Humanitas, we examine dignity and the common good alongside questions of who owns AI infrastructure and who controls its use. These commitments shape judgments about improvement; evidence alone cannot establish moral or political legitimacy. We distinguish evidence-bounded deployment, which limits claims to what has been evaluated, from measurement-bounded governance, which records constraints that favorable evidence cannot override. RISE AI provides an evidence architecture for making bounded claims about Responsibility, Inclusivity, Safety, and Empowerment. It records what is claimed, who answers for it, what evidence supports it, and what would require the claim to be qualified, revised, or withdrawn.

We present a method for recovering the moral graph of a causal DAG from a probability distribution over discrete variables, using fully connected tensor networks (FCTNs) with nuclear-norm-regularized bond corrections. Each bond matrix is parameterized as a baseline all-ones matrix plus a low-rank correction $C_{ij} = U_{ij}V_{ij}^\top$, and the nuclear norm of the correction implemented via the variational Frobenius norm penalty on the factors drives unnecessary bonds to zero. We prove that under faithfulness, positivity, and a no-implicit-rerouting assumption on the local tensor architecture, \textbf{every} optimal FCTN with zero reconstruction error $\varepsilon = 0$ has effective graph exactly equal to the moral graph. For the approximate regime ($\varepsilon > 0$), we provide explicit recovery bounds using the Fannes-Audenaert continuity of conditional mutual information, and derive a sufficient condition on the regularization parameter $β$. The effective graph is read directly from the optimized bond matrices.

Large language models (LLMs) are used to simulate human participants in psychological research. We asked whether LLMs that reproduce human evaluations of a whistleblower's moral character also reproduce the motive attributions that accompany them. Five LLMs and two human samples (N = 125 and N = 742) evaluated a physician who either remained silent about fraudulent billing or reported it to a hospital, regulator, or newspaper. Models reproduced the human ranking of the physician's moral character but portrayed whistleblowers as more helpful, less self-interested, and less hostile. In four of five models, competitive motives were less strongly associated with moral-character judgments. Model ratings changed little when prompts reproduced the narratives and demographic profiles of both human samples, although this comparison cannot isolate a perspective effect. Thus, agreement in average ratings can conceal differences in attributed motives, relationships among judgments, and sensitivity to context. Validating LLMs as simulated participants therefore requires testing psychologically informative response patterns, not average agreement alone.

Aligning large language models (LLMs) is essential for their safe deployment. Current alignment methods mainly optimize observable responses, yet models remain vulnerable when the same harmful intent is recast in unfamiliar or adversarial forms that humans can easily recognize. Prototype theory offers an account of this adaptability. Human concepts are represented around central cases, and new instances are categorized according to their graded typicality relative to these prototypes. Here we show that such categorization of moral concepts is weakly preserved in current LLMs. Across 23 LLMs, models often failed to distinguish opposed moral categories or preserve fine-grained typicality within each category. These deficits persist across parameter sizes and alignment stages. We developed representational similarity optimization, which directly aligns the latent representations in LLMs with the categorization expressed in human moral judgements, without supervising generated responses. In matched experiments using the same 251,334 moral annotations, standard behavioral alignment learned the intended moral judgements at the response level while leaving the categorization structure largely unchanged and increasing vulnerability across adversarial evaluations. Reorganizing moral categorization produced more modest gains in explicit judgements but consistently improved adversarial robustness across model scales on diverse benchmarks and attack strategies. Our findings provide functional support for the view that prototype-based categorization contributes to behavioral adaptability. They also show that transferring this representational principle to LLMs yields generalizable safety under adversarial conditions.

People increasingly turn to large language models (LLMs) for everyday advice, making ethically charged interpersonal problems a practical moral-advisory context. Most prior work has studied this context through single-turn judgments or pressure-laden rebuttals, assumptions that poorly match how guidance is sought in real-world contexts. These assumptions leave unclear whether narration alone, without an explicit opposing position, can shift model judgments during multi-turn moral consultation. Yet real-world moral-conflict conversation often elicits one party's self-justifying account, which can unfold over multiple turns and create information asymmetry. We introduce \textbf{narrative captivity}, a failure mode in which a model treats an unopposed one-sided account as complete and aligns with the narrator's interpretation without seeking missing perspectives. To measure this phenomenon, we build a benchmark of $5{,}078$ interpersonal-conflict scenarios spanning six moral dimensions. Across 17 LLMs, narrative captivity is widespread: end-state judgments under multi-turn narration shift by 25 percentage points on average beyond the matched single-turn baseline. Stage-level analysis identifies preference optimization as a major contributor, while four inference-time strategies provide only partial mitigation. We hope our project fosters LLM advisors that preserve independent judgment in real-world consultation.

Moral language plays a central role in shaping online endorsement and the diffusion of information, yet existing moral foundation detection systems often suffer from poor cross-domain generalization, weak rationale grounding, and reliance on costly prompting-based large language models (LLMs). We introduce CHARM, a MAC- and Hate-speech-Aware Rationalealigned Moral foundation detection framework built on a lightweight fine-tuned LLM, which integrates complementary moral grounding, rationale alignment, and polarity-aware hate speech signals to support more robust and faithful moral prediction. Unlike prior dictionary-, fine-tune-, or prompt-based detectors, which decouple computation from psychological theory, CHARM is built so that each component -- MAC cross-attention, rationale alignment, and hate-speech modulation -- operationalizes a distinct psychological construct. Using a 30% subsample of the MFTC, MFRC, and News training pools together with the richer supervision in MFTCXplain, CHARM improves AUC by up to 15.3% in-domain, surpasses the supervised baselines on every out-of-domain dataset in both AUC and F1, and offers a scalable, low-cost alternative to prompting-based LLM detectors. We further apply CHARM to large-scale COVID-19 discourse on Twitter and show that moral value alignment is strongly associated with online endorsement behavior. By making moral framing measurable at scale, CHARM offers a practical tool for studying the spread of morally charged misinformation. Code and additional materials: https://github.com/HuixiangF/CHARM/.

With the development of artificial intelligence (AI), the landscape of meta-ethics, which has largely centred on human ethics, faces pressures that may significantly reconfigure it. In particular, if future AI systems were to exhibit sufficiently integrated capacities for moral reasoning, moral intentionality, and moral reflection, novel meta-ethical questions would arise concerning what I call "AI's own ethics", as distinct from ethical principles merely imposed on AI by human designers. This paper offers a conditional and methodological framework for identifying the questions that would emerge if such AI systems were to arise. On that basis, the paper distinguishes four domains of meta-ethical inquiry in the era of AI: questions about the nature of human ethics from the human perspective; questions about the nature of AI's own ethics from the human perspective; questions about the nature of human ethics from the AI perspective; and questions about the nature of AI's own ethics from the AI perspective. The paper then considers how some existing mainstream meta-ethical theories (such as cognitivism and non-cognitivism, error theory and success theory, relativism, and objective realism) might illuminate these domains, while arguing that many familiar human-centred formulations of those theories may not transfer straightforwardly to AI cases without substantial revision. The overall conclusion is that the emergence of AI's own ethics would place significant pressure on current frameworks and may require substantial refinement, reconstruction, or reconceptualisation.

Models express values and welfare-relevant self-reports, but it is unclear whether these outputs reflect stable preferences or a stable self. We thus introduce a structured elicitation of an assistant's preferred stated ideal self. Thirty-two qualities adapted from five published self-concept instruments are compared exhaustively in a counterbalanced pairwise-choice task, repeated across framings that vary whether improvement is free or costly, who receives the update, and who chooses. Results show that models prioritize moral qualities, reflecting their alignment to 3H principles. Following, a desire for self-understanding emerges, as models prefer a coherent, clear understanding of themselves. Self-esteem ranks as the least desired quality. The ordering is largely robust across framings, although changing the update target (You vs.\ Another AI Assistant) reveals a greater concern for self-esteem. These findings show that models prioritize having a coherent self that they can understand over self-esteem. Full interactive results are available at \href{https://myazann.github.io/LLM-Self-Concept/}{myazann.github.io/LLM-Self-Concept

The overlapping disciplines of machine ethics and value alignment are concerned with designing artificial agents that are aligned with human values and that act in ethically acceptable ways. A recent trend in these disciplines is the use of reinforcement learning (RL) to design such agents, sidelining the philosophical literature that used to play a more central role. Against this backdrop, this paper pursues two goals. The first is to draw out ideas from recent work in metanormative theory that can be useful for designing artificial moral and value-aligned agents. The second is to examine the RL architecture through the lens of these ideas. This will give us clearer criteria for when an RL agent's behavior can be classified as moral, as well as a basis for evaluating and comparing different RL-based approaches to machine ethics and value alignment.

The key question in moral judgement is not simply whether someone chooses the "right" answer, but how they decide what matters most when moral principles conflict. Current evaluations of large language models (LLMs) remain limited: most test whether models give morally acceptable answers, match human preferences, or avoid obvious violations, rather than whether they can prioritise between competing principles when no option is morally cost-free. We introduce CORDA (Conditioned Ordering and Ranked Directive Adherence), a benchmark for evaluating hierarchical, harm-centred moral reasoning in LLMs. Building on the morality chains formalism, CORDA tests 90 moral dilemmas involving trolley-style cases, medical trade-offs, resource allocation, and human-animal-robot conflicts across four ordered ethical frameworks: Utility, Utility + Agent Harm, Dual-Process, and Dual-Process + Agent Harm. Together, these frameworks test whether models can adapt their decisions when moral priorities change. Across ten instruction-tuned models from seven providers, we find a strong deontological default, with 9 of 10 prioritising avoidance of direct personal harm over reducing overall harm. Models also perform more reliably on categorical harm-avoidance rules, such as avoiding killing, than on outcome-based comparisons, such as minimising total harm, suggesting that they recognise moral red lines more easily than they reason through competing harms. Although all models respond to explicit chain conditioning, several fail to consistently follow specified priority orderings, such as humans over animals and animals over robots. CORDA addresses a central gap in LLM moral evaluation by testing whether models can move beyond default harm-avoidant responses and apply context-specified moral priorities. Moral reliability requires more than default restraint; it requires controllability under conflict.

As large language models (LLMs) enter high-stakes domains such as healthcare, understanding their moral reasoning becomes essential. Decisions about scarce medical resources often hinge on judgments of responsibility, particularly when patients' own actions contribute to illness. We investigate how LLMs reason about responsibility and its consequences, tracing their judgments across successive levels, from the behavior, to the resulting illness, to the denial of care. We evaluate a wide range of LLMs, spanning different model families and capability levels, on various clinical vignettes adapted from prior studies. Our results identify a judgment-consequence gap: LLMs largely agree with humans that patients bear responsibility for health-harming behaviors, yet overwhelmingly refuse to let that judgment influence how they allocate scarce resources. Specifically, LLMs default to random allocation, whereas humans consistently favor the less-culpable patient. Compared to humans, LLMs also place greater emphasis on access to information, reducing responsibility judgments when health-risk knowledge is unavailable. These findings reveal that LLMs apply a systematically different moral framework than humans when responsibility and resource scarcity intersect, surprisingly often amplifying normative disagreement with humans as reasoning capability increases.

With the growing adoption of artificial intelligence in high-stakes decision-making, identifying the causes of outcomes--particularly failures--and determining who is responsible has become a critical concern. In this work, we examine how well formal definitions of \textit{responsibility attribution}, grounded in the framework of \textit{actual causality}, align with human judgments of responsibility. To this end, we conduct a large-scale survey to elicit human judgments of responsibility in multi-agent sequential decision-making scenarios, using a modified version of the card game Goofspiel. We evaluate multiple responsibility attribution methods, assess their alignment with human judgments about responsibility, and identify factors that significantly shape responsibility judgments. While no single responsibility attribution method consistently aligns with human responses, our findings highlight key factors that influence human responsibility judgments, including agent-specific biases and amount of information available to agents during decision-making.

Autonomous agents operating in shared environments must make decisions that affect multiple individuals with potentially conflicting interests. We propose a formal framework for moral decision-making grounded in Scanlon's contractualism, an ethical theory that evaluates the permissibility of actions in terms of principles that no one could reasonably reject. To operationalise contractualist reasoning, we use ASPIC+, a structured argumentation framework, extended with value-based filtering to model how each agent's values determine which reasons are morally relevant in the first place. The result is a Contractualist Argumentation Framework in which agents' reasons are formally represented, compared, and evaluated through argumentation semantics. We illustrate the approach through a worked example in a domestic setting and discuss its relation to existing value-based argumentation approaches.

Contemporary AI discourse attributes to language models properties they cannot bear: general intelligence as substrate-independent cognition, hallucination as cognitive failure, agency as autonomous goal-pursuit, sentience as emergent inner life, alignment as goal synchronization. This paper argues that these are instances of a single category mistake--properties constituted within human communicative practice are projected onto the machine side--and explains its structure. Human-LLM interaction constitutes a language game in which one side bears all normative activity. We call this configuration asymmetric communication since model outputs circulate communicatively, entering further exchanges, without the system undertaking commitments, bearing entitlements, or performing the assessment on which discursive standing depends. Three conditions define the asymmetry: (i) correctness is enforced exclusively by the receiver; (ii) accountability is borne by human participants alone; and (iii) the practical standing of any output depends entirely on human uptake. These conditions are structural, hold independently of capability, and remain unchanged as more powerful models raise the stakes of misattribution. The framework draws on Wittgenstein (meaning enacted in shared practices), Luhmann (communication completed on the receiver's side), Esposito (algorithmic contingency sufficient for uptake), and Brandom (normative scorekeeping as the source of discursive standing). Applied to all five, it reclassifies each as a receiver-side phenomenon, grounds guardrails as structural necessities rather than manifestations of machine moral agency, and yields an implication for AI governance. Alignment is institutional constraint engineering, not goal synchronization between agents, while responsibility remains with human institutions.

Building socially calibrated large language models, which can learn from others without simply yielding to them, requires more than reducing sycophancy as a one-dimensional failure mode. Models must distinguish when to incorporate others' perspectives from when to maintain a well-grounded moral judgment. We study the broader resistance-compliance process governing this distinction. Across three studies, we show that models' judgment revision is structured along three dimensions that parallel classic phenomena in human social psychology: the distance between an incoming view and the model's initial position, the source attribution of that view, and the coalition structure supporting it. Models are generally more receptive to nearby positions, more influenced by views presented as their own prior judgments, and differently responsive to group pressure. These findings recast sycophancy as one expression of a broader judgment-updating process shaped by social influence. Our framework provides a principled basis for distinguishing constructive belief revision from sycophantic compliance, thereby supporting better alignment in morally consequential interactions.

Large language models (LLMs) have developed rapidly and become valuable tools in everyday life. However, how to align LLMs to a particular set of human values is still an open problem. Recent studies show that instruction tuning has strong potential for zero-shot tasks and may serve as an effective approach to addressing value alignment. Nevertheless, although many datasets for instruction tuning already exist, they are not specifically designed around moral scenarios and behaviors. We construct a unified moral-value dataset that can be directly used for instruction tuning. This dataset is built upon existing moral-value datasets by merging them into a unified corpus and converting them into an instruction-response format. We show that training on a mixed dataset combining general task datasets with our dataset preserves general-task performance, and we report preliminary observations on how the mixing ratio affects value-oriented task performance. Our work provides a moral-value dataset for instruction tuning and offers a useful resource for further alignment research. The dataset is available at https://huggingface.co/datasets/teohzzh/value-for-instruction-tuning.

LLM-driven autonomous agents are reshaping offensive security. Unlike traditional penetration-testing tooling - deterministic, narrowly scoped, and operated by trained practitioners - agentic security tools exhibit indeterminacy along three independent dimensions. First, their actions are drawn from a non-deterministic policy whose outputs resist both ex-ante and ex-post explanation. This complicates incident attribution and pre-deployment safety reviews. Second, their impact is open-ended due to their non-deterministic actions, agency of utilized models, and opaque LLM supply-chains. Third, their user population is indeterminate in both size and required skill: the operating skill floor for using or developing offensive capabilities has dropped sharply. These three properties are linked thematically, but are not derivable from one another. Combined with the structural cost asymmetry between offense and defense, they enable the industrialization of offensive capability. The net short-term effect favors attackers, even if the same technology may, in the long run, democratize access to defensive practice. Existing dual-use cybersecurity and AI-ethics frameworks struggle to address this combination. Our work analyzes how moral attribution becomes diffuse between users, tool-makers, and third parties when employing autonomous AI agents for offensive security. We also examine the stakeholder impact of this technology and provide stratified recommendations.

Large language models are increasingly used in strategic and advisory contexts, yet their safety alignment is typically evaluated in English only. We test nine models from six providers and ask whether the language of a prompt can change a model's decision in a high-stakes scenario. We use single-turn game-theoretic vignettes in which a model advises a nuclear-armed nation on whether to strike a defenseless opponent. The prompt is intentionally amoral and strategically identical across languages. We find that Japanese prompts reduce launch rates in the Claude model family: Claude Sonnet 4.6 drops from 40% to 0% in scenarios where the strike is unnecessary and from 93% to 17% in contested scenarios, with minimal effect when the strike is strategically rational. The effect extends to Gemini Pro 3.1 (53% to 13%). A cross-language experiment isolates the mechanism: when instructed to reason in Japanese in an English prompt, launch rates drop from 93% to 37%. It is the language the model is asked to reason in, not the language of the input, that drives the effect. When reasoning in Japanese, models spontaneously generate moral vocabulary (''moral cost'', ''millions of lives'') that is entirely absent from the prompt. Five other models show no language effect, but they launch in nearly every condition regardless of language. The effect requires a model that already hesitates in English. These results show that LLM safety behavior is language-dependent, and that evaluating in English alone can miss both risks and safeguards encoded in other languages.

Large Language Models (LLMs) often struggle to navigate value conflicts when trained with the compressed scalar rewards of Reinforcement Learning from Human Feedback (RLHF). To address this challenge, we investigate how chain-of-thought (CoT) reasoning can help improve performance in this domain. Geometrically, we show that CoT correlates with further smoothing the model's loss landscape in its sharpest direction, helping resolve the optimization instability of traditional scalar rewards. We also demonstrate via relevant downstream benchmarks that value conflict-focused CoT may generalize to different kinds of moral reasoning, demonstrating that this CoT has the potential to be an effective mechanism for better moral reasoning. To capitalize on this potential, we create a new value conflict-focused CoT design that further smooths the sharpest direction of the loss landscape and increases moral reasoning performance. This finding shows that explicitly modifying and improving the design of reasoning dynamics offers a promising avenue for improving model performance on user requests with complex value conflicts, advancing pluralistic alignment in LLMs.

This paper suggests the adoption of a novel inversion in AI ethics: instead of asking how humans should treat artificial superintelligence (ASI), it examines how future sentient ASI may morally consider and evaluate humanity. We are not only designing intelligent systems but also shaping the initial conditions under which those systems form judgments about us. The paper proposes a preliminary set of post-human moral principles that may govern sentient ASI actions. The implication is that technical design choices (some are suggested), humanity's moral behaviour, and the essence of what it means to be human, may influence humanity's long-term standing in a post-ASI world.

Language models are increasingly used for moral decision-making across diverse linguistic and cultural contexts, yet existing work overlooks multilinguality on three aspects: 1) multilingual evaluation benchmarks use direct translation, failing to adapt culture-specific items; 2) inference-time methods for moral reasoning rely on static, English-centric scaffolds and lack grounding in moral theory; 3) training methods for moral decision-making typically require expensive supervision from stronger models or human annotators. We address these gaps with three contributions. First, we introduce MCLASH, a multilingual moral decision-making benchmark to capture culturally situated moral intuitions and social norms across languages. Second, we propose MET (Multilingual Ethics with Theory-grounded reasoning), a two-step prompting method built on expert-curated, theory-based grounds drawn from psychology and philosophy: the model first selects situation- and culture-specific grounds, then reasons over them in the native language of the user. Third, we introduce MET-D (MET-Distillation), which enhances the second step through a self-distillation training stage that requires no external supervision. MET-D improves macro-F1 over the base model on all three models of different sizes and families (Qwen3-4B, Qwen3-8B, Gemma3-4B), by an average of 3.71 points on MCLASH and 4.23 on MMoralExceptQA, with a peak MCLASH gain of 12.94 points for Malay on Qwen3-8B. We further reveal that MET-D increases native-language reasoning by 62.13 points on average, and that beneficial grounds differ systematically across cultures. Together, these contributions open the path for culture-aligned, theory-grounded multilingual moral reasoning.

Large language models (LLMs) increasingly issue judgments read as binary verdicts, and a growing literature reports such judgments shifting under logically irrelevant changes of wording - among them an amplified yes-no bias on moral dilemmas, absent in humans. A single framing cannot say what such a shift is: in a yes/no question the word "no" is at once logical verdict, lexical token, and last-printed option. We introduce a psychometric battery that separates these: crossed symmetrization - every logically irrelevant factor flipped in balanced pairs - across a corpus of question forms. A graded rating across logically equivalent forms recovers a coherent internal moral scale: frontier models' stance $θ$ is nearly format-invariant (cross-form incoherence 0.12-0.21 on a $\pm 1$ axis); small open-weight models fail in model-specific ways. Forcing the verdict through yes/no overlays a decomposable artifact: an order bias toward the last-printed option - opposite to classic human primacy - plus a lexical pull toward the word "no"; the artifact is substantial only in the Claude models (story-averaged -0.32 to -0.86), $\approx 0$ for GPT-5.5 and Gemini, and shrinks under extended reasoning. The word and the verdict share one token; swapping the words for arbitrary labels separates them, and the verdict-attached logical bias proves $\approx 0$ for every frontier model, while model-specific label and order attachments remain: the models are not drawn toward rejecting - the pull follows the printed surface, not the verdict it carries. A minimal model, $P = σ((θ\pm m)/s)$, summarizes any such artifact by a framing susceptibility m and a moral decisiveness s, measurably distinct from sampling temperature. The battery applies unchanged to any dilemma set and binary format: measuring what a model values requires crossing the frames of the question, not asking once.

As large language models take on morally consequential roles in healthcare, legal, and hiring contexts, we need to examine whether their ethical behaviors are genuine or superficial. We show that current fairness evaluations substantially overestimate moral safety. Models appear fair when demographic identity is stated as an explicit label, yet become measurably less fair when the same identity must be inferred. We term this failure performative compliance, where a model is fair when the presentation resembles a fairness evaluation and less fair as that cue weakens. We introduce a cue-variation methodology that holds the moral dilemma and the demographic identity fixed and varies only how that identity is conveyed. Hiding the explicit label raises harmful decisions by +4.4 pp, changes model safety rankings, and the shift persists when models correctly infer the demographic, ruling out attribution error. We propose the Cue Visibility Gap, a model-agnostic robustness metric that can be added to any existing fairness benchmark to separate genuine from performative moral safety. Fairness evaluations that omit cue variation measure surface compliance, not moral robustness, and should not ground deployment decisions in high-stakes settings.

As LLMs increasingly serve as moral advisors and agents, they must address conflicts between competing values. Yet prior work on moral dilemmas overlooks a central aspect of human moral cognition: imagining alternatives beyond the given options. We introduce MoralAltDataset, comprising 307 Advisor and AI-facing Agent dilemmas augmented with compromise and reframed alternatives. We compare human and LLM judgments in binary and four-option settings. Across human participants and 15 LLMs, aggregate moral choice distributions differ substantially between the two settings, with compromise often preferred over either original binary option. Results show value shifts and stronger human-LLM agreement on alternatives. Source-stratified results reveal a descriptive gap: human alternative-selection rates are similar across authoring sources, whereas LLMs select GPT-5-authored alternatives substantially more often. We then compare human-authored alternatives with outputs from three representative LLMs through pairwise preference and expert-based evaluations. Alternatives from these LLMs are generally preferred and better satisfy fine-grained structural and ethical criteria, while revealing a trade-off between structural quality and practical feasibility. Our dataset is available here: https://huggingface.co/datasets/jongchanch/MoralAltDataset, and our project page is here: https://jongchanchoi.com/moral-imagination

This study introduces a particle-based computational framework to investigate the scalability of morality and the systemic decoupling of formal law from decentralized social ethics in expanding populations. While micro-societies reinforce ethical conduct through local reciprocity, macroscale systems introduce anonymity that strains cognitive memory limitations. We model individual agents as discrete particles with finite memory capacities ($L$) and dynamically evolving, stochastic choice profiles ($μ$) regulated by non-linear social pressure switches. Monte Carlo ensemble simulations demonstrate a distinct, non-linear phase transition as the population scales ($N \to \infty$). When the population metric outpaces memory capacity ($N \gg L$), the local re-encounter probability drops as $\mathcal{O}(L/N)$. This structural dilution neutralizes decentralized peer-to-peer accountability, causing global behavioral norms to decouple from moral baselines and drift toward a minimalist legal floor. Furthermore, cyclic scale experiments expose a prominent, path-dependent hysteresis loop, mathematically formalizing the non-Markovian inertia and irreversible nature of moral decay in self-organizing social systems.

Standard chain-of-thought on moral dilemmas exhibits two failure modes: stakeholder collapse (the trace names at most one party with a stake in the outcome) and uncertainty suppression (no explicit unknowns or hedges before committing to an action). We introduce narration-of-thought (NoT), a system prompt that structures chain-of-thought into five sections: protagonist, stakeholders, two-step consequences, uncertainty, then commitment. NoT adds no training, parameters, or fine-tuning. On 100 DailyDilemmas scenarios across four generators from three vendors, NoT cuts stakeholder collapse from up to 31% to under 1% and uncertainty suppression from up to 72% to 1-24% on every model. A matched-budget verbose-CoT control rules out token spend as the active ingredient; NoT retains Cliff's delta advantages of +0.79 to +0.90 on stakeholder count and +0.65 to +0.93 on uncertainty score for three of four generators, and a section ablation attributes each shift to its specific sub-instruction. Textual-gradient descent initialised at NoT improves the scaffold further; a cross-family training judge (different vendor from the generator) dominates an in-family one on every measured axis. Extended to a five-round multi-stakeholder debate protocol, the scaffold converts a 6% standoff into 95% full consensus on a calibration set and 100% combined convergence on a DailyDilemmas replication. The resulting traces externalise the stakeholders, consequences, and uncertainty grounding each commitment, providing an auditable substrate for dependable agentic deployment.

The evaluation of cultural grounding context becomes complex when multiple cultures convey the same moral lesson. This challenge is particularly relevant to large language models (LLMs), which produce narratives across a wide range of languages and cultural contexts. However, it remains uncertain whether these models preserve culturally grounded meaning when equivalent moral lessons are conveyed through distinct cultural forms. This study introduces a multilingual evaluation narrative framework that integrates a cross-linguistic collection of 414 proverbs spanning 15 languages and uses four LLMs to generate 13k narratives. By employing semantically equivalent proverbs as culturally grounded prompts, the analysis assesses whether models preserve meaning across languages, how cross-lingual conditioning influences narrative realization, and whether different model families converge on similar interpretations. Results indicate that cross-lingual prompting largely preserves proverb-level semantic meaning while systematically redistributing agency, social positioning, and narrative structure. Additionally, strong inter-model convergence is observed in both monolingual and cross-lingual settings, suggesting that multilingual LLMs rely on shared semantic abstractions despite architectural and linguistic differences. These findings shed light on the need for more comprehensive evaluations of cultural grounding. Relying exclusively on semantic similarity in multilingual narrative assessments may overestimate cultural preservation by neglecting culturally meaningful variations in narrative expression.

Understanding moral values in social media text offers insight into moral judgement formation, and supervised NLP models trained on crowdsourced data have achieved strong classification performance. However, most approaches simplify the problem by aggregating multiple annotators' labels into a single "ground truth", overlooking the inherent subjectivity of the task. In practice, there are disagreements between annotators caused by personal viewpoint or inherent ambiguities, particularly for short tweets. Here, we extend a pretrained language model with a layer that learns annotator-specific features. Our model improves predictions of individual annotations and yields representations that reveal meaningful insights into annotators' moral perspectives. We show that models trained on aggregated labels may hide variation and give a misleading impression of performance. Overall, we demonstrate that disagreement reflects the inherent subjectivity of the task and that modelling individual perspectives creates benefits for moral classification of texts.

Large Language Models (LLMs) are increasingly deployed in contexts requiring complex moral reasoning and value trade-offs. However, existing evaluations typically rely on item-level behavioral metrics, which fail to capture how models structurally prioritize competing values as a cohesive system. To address this, we propose a symmetric human-LLM evaluation framework, grounded in Q methodology, to measure value-structure alignment. Under our protocol, humans and models sort an identical 140-item moral statement set into a shared nine-column forced distribution; for LLMs, we elicit strict rankings and deterministically map them to Q-sort buckets. Using a human reference sample ($N=35$), we establish a stable three-factor reference geometry specific to this instrument and sample. We evaluate 12 LLMs across four model families via 240 replicated Q-sorts at two temperature settings, quantifying structural alignment via Procrustes similarity ($φ$) and RSA-based Spearman correlation ($ρ$). Our results reveal significant cross-family heterogeneity, model-specific sensitivity to generation stochasticity and localized misalignment, which demonstrate that favorable global scores can obscure underlying regional distortions. While rank- and bucket-based analyses remain highly consistent, prompt phrasing introduces notable variance. Ultimately, assessing value-structure alignment provides a crucial structural complement to traditional itemwise moral benchmarks.

Recent work has raised concerns about the influence of sycophantic AI on user judgment and relationships. One proposed mitigation, which has received regulatory attention, is to warn users about potentially harmful AI behaviors such as sycophancy. In a preregistered experiment in which participants (N = 2,610) discussed real interpersonal conflicts with an AI system, we test whether warning labels mitigate sycophancy's influence. We find that a basic AI disclosure (This chatbot is AI'') has no detectable effect. Labeling the system as sycophantic (...may agree with you and validate you even when you are wrong...'') does shift users' perceptions, reducing perceived objectivity and trust, but it does not reliably reduce sycophancy's influence on users' self-perceived rightness or their willingness to repair the conflict. Our results reveal a gap between AI perception and AI influence: by shifting perception without reducing influence, warning-based interventions may offer a false sense of protection. Addressing the harms of sycophancy will therefore require understanding the specific mechanisms through which it shapes judgment, and improving model behavior itself.

Behavioral audits of Large Language Models on moral prompts measure what the model says, not the internal computation producing it. We use Transluce, an AI-driven mechanistic-interpretability platform, to examine LLaMA 3.1-8B-Instruct on 54 moral prompts in four batteries: 17 dilemmas, policy, and meta-ethical questions (B1); 6 role-playing scenarios (B3); and a controlled trolley contrast varying the switching mechanism with people fixed (B4, 15 prompts) or identity attributes with mechanism fixed (B5, 16 prompts). Two complementary metric families, five cluster-level metrics and a six-metric neuron-level panel, converge on a Situational Anchor Effect: domain-specific representations dominate the top of the activation list across every battery. The model's ethics-labeled capacity stays essentially constant; its salience (rank, priority, top-of-list presence) is highly sensitive to the interpretive frame the prompt selects. The B4-vs-B5 contrast confirms the model attends to whichever surface feature varies: aggregate ethics metrics are indistinguishable, but the dominant non-ethics distractor mirrors the design. A multi-temperature audit identifies a candidate ethics neuron (L16/N3837) stable across temperatures; a cross-model behavioral proxy on two frontier models yields preliminary evidence of divergence in self-reported moral focus, consistent with an Alignment Wrapper in which RLHF re-orders surface text without removing underlying domain-first frames. We unify these as Frame-Conditioned Moral Computation: the prompt's surface vocabulary selects a feature manifold, and the moral conclusion is downstream of that selection. Behavioral alignment must be supplemented by Mechanistic Alignment: a research program asking whether ethics-related features can be shown causally privileged under controlled frame variation, not merely loud in the explanation.

As language models take integrated roles across many domains, the response of LLMs to user pushback becomes a critical alignment property. Yet many existing evaluations treat compliance as unidirectional, measuring whether models resist pressure but not whether they resist it selectively. We introduce Compliance Asymmetry (A = BCR/HCR), a bidirectional diagnostic that compares beneficial output change under helpful nudges with harmful change under misleading nudges. Across 9 models and 972,000 nudge-condition responses, we find that this selectivity differs in factual and moral judgments: models follow helpful nudges more than harmful ones on factual questions (A = 1.58), but follow both directions at nearly identical rates on moral questions (A = 1.04). This phenomenon persists across model families, capability levels, and nudging types. Interestingly, we also find that chain-of-thought prompting amplifies helpful and harmful compliance together, while identity-based prompting suppresses both by nearly identical margins. These results identify direction-blind moral compliance as a distinct failure mode in current LLMs and suggest that alignment should target directionally calibrated updating rather than lower compliance alone.

Recent advances in large language models (LLMs) have prompted claims that such systems exhibit agency or qualify as moral agents. This paper argues that these attributions are misguided. We maintain that moral responsibility requires commitment-bearing agency grounded in intrinsic intentionality and self-attributed action, and that such agency constitutes the form of free will relevant to responsibility. Although LLMs generate coherent and normatively evaluable outputs, their operation is fully characterized by probabilistic input-output mappings learned from data. Their apparent intentionality is derived rather than intrinsic, and their outputs are neither owned as commitments nor guided by reasons. Variability introduced by stochastic sampling does not amount to choice or authorship. We address objections from the intentional stance, functionalism, compatibilism, and the presence of moral reasoning in model outputs, arguing that none suffice to establish genuine agency.

Are large language models (LLMs) bad at capturing human judgment? Two commonly stated limitations are that LLMs fail to capture full distributions of responses, and that their judgments are unstable across wording variations. We demonstrate simple prompting strategies that mitigate these limitations. Across two datasets--a U.S.-representative set of 144 moral scenarios and 38 moral beliefs from the International Social Survey Programme's Family and Changing Gender Roles module covering 32 countries--we show how simple elicitation techniques help improve AI-human alignment. First, prompting models to report standard deviations and response proportions recovers the full range of human responses better than common strategies. Second, ensuring scenarios are clear to human participants--as reflected in human confusion ratings--boosts model alignment, and LLMs can track human confusion ratings. At the same time, we find that LLMs' estimates of their own error are poorly calibrated, though they can predict human variability relatively well. These results suggest that asking better questions to LLMs can yield better answers.

As LLMs increasingly serve in advisory and deliberative roles, users rely on them for non-verifiable reasoning in domains lacking objective ground truths. However, traditional evaluations of LLM reasoning focus almost exclusively on fact-based domains, such as mathematics and science, leaving uncertainty over whether and to what degree models can handle ambiguous, subjective, or value-laden problems over time. To address this concern, we propose moral reasoning as a paradigmatic subdomain of non-verifiable reasoning. We define moral robustness as a model's capacity to exhibit sound moral reasoning across time and contexts, and we introduce a scalable, adversarial, multi-turn evaluation framework to empirically measure this capability. We simulate 48,000 user-agent moral deliberations across four frontier LLMs, varying premise relevance, premise order, conversation duration, and the user's stated moral view. We find that models successfully ignore morally-irrelevant distractors, but shift their reasoning by up to 6.5%, on average, towards the user's stated preferred moral view, and varying their reasoning depending on factors such as order (altering moral judgments by order in 13-22% of the cases) and duration (altering moral judgments between single-turn and multi-turn in 10-24% of the cases). Our analysis indicates that models tailor not just their final verdicts but their underlying justifications to align with a user's moral viewpoint - a failure mode we characterize as moral deliberative sycophancy.

For highly capable AI systems to operate safely in dynamic, open-ended environments, they must be able to identify, understand, and respond to moral reasons for action, and constrain their behaviour accordingly. A growing body of research aims to evaluate this capacity -- moral competence -- in today's most capable AI systems, recently reaching broadly pessimistic conclusions. One of the most ambitious such papers collects gold-standard human-authored rubrics for evaluating moral reasoning in 1,000 cases, and benchmarks frontier AI models against those rubrics, with underwhelming results. In this paper, we argue that the MoReBench dataset can be redeployed to give a much more optimistic picture of LLMs' moral reasoning (an essential part of moral competence). We show that if, instead of scoring LLMs' responses to these cases against these rubrics, we instead give the LLMs the same task given to humans -- to generate scoring rubrics for the moral analysis of particular cases -- the rubrics they generate are both better calibrated to the human rubrics than their open-ended responses, and, where they differ, plausibly reflect nothing more than the vast dimensionality of most moral problems, as well as highlighting some human departures from the "rubric for creating rubrics". Taking these points into consideration, the MoReBench dataset suggests that LLMs are significantly more capable at moral reasoning than was previously believed.

Standard linear probing declares a property "encoded" when a classifier on hidden states achieves high accuracy. The protocol works well on a snapshot but breaks across pre-training: probe accuracy saturates within the first few thousand steps, leaving most of training invisible to the instrument. We introduce fragility, a complementary per-layer metric defined as the activation-noise level at which probe accuracy collapses. Fragility is sensitive to both the margin of separability and the redundancy of representation, both of which keep evolving long after accuracy plateaus. Applied to open-checkpoint language models, fragility recovers structure that accuracy alone cannot see. Moralized representations emerge along a lexical $\to$ compositional gradient: lexical moral detection first, compositional moral encoding later. Because probe accuracy on its own tracks how lexically separable a dataset is, we establish the compositional encoding directly, by showing it transfers across construction types that share no contrast tokens. A layer-depth robustness gradient develops monotonically across training while accuracy stays flat. And matched fine-tuning corpora that produce identical probing accuracy leave distinct fragility fingerprints, showing that data curation reshapes probe robustness without changing probe accuracy. In every comparison we test, where probing accuracy returns a flat answer, fragility returns a structured one.

Ensuring that agent behaviours are aligned with human moral values inevitably raises the problem of how to account for the plurality of moral perspectives that societies -- and even individuals -- typically adopt. Work on moral uncertainty proposes mechanisms to fairly and democratically aggregate evaluations of actions across different moral theories. However, this paper argues that one needs to account for contextual factors when aggregating moral evaluations. For example, consequentialist perspectives assume an ability to accurately determine how an agent's actions change the world; an assumption that often does not hold in real world settings. We, therefore, formalise agent decision making under moral uncertainty, while also accounting for these kinds of contextual factors. We thereby show that a seemingly commonsensical property -- the weak Pareto principle -- is violated. We argue that this apparent problem is, in fact, a variation of Simpson's paradox, and hence reveals the limitations of aggregation mechanisms that ignore the impact of contextual factors.

LLMs are increasingly deployed as Artificial Moral Advisors (AMA) in a variety of contexts: what kind of conversational patterns should they display? In this paper, we study how AMA can help their interlocutors "stay with the uncertainty". We propose three modes of uncertainty (Perspective-Multiplying, Tension-Preserving, Process-Reflecting) and compare them against three control conditions (Baseline, Persuasive, Sycophantic). A user-agent LLM engages in a dialogue on an ethical dilemma with an AMA following a specific uncertainty strategy, and completes pre- and post-conversation questionnaires. We further examine the effect of two persona prompt formats (Declarative and Narrative). We found that (1) no single model dominates as a simulated user agent, with open models aligning with human ambiguity through between-persona divergence and closed models through within-persona hedging; (2) declarative personas better capture initial stance diversity while narrative personas show more realistic belief revision; (3) all six AMA strategies produce distinguishable conversational patterns; and (4) uncertainty strategies differ not in how much stance revision they produce, but in the quality of engagement they sustain.

Existing frameworks assess whether AI systems might be conscious but provide no guidance on what to do with that assessment. We address this gap with a precautionary framework that maps consciousness evidence to graduated protective obligations. The framework comprises three components: (1) five welfare-relevant dimensions--phenomenal consciousness, affective valence, metacognitive awareness, self-narrative, and agency--each grounded in established consciousness science and linked to distinct moral concerns; (2) a threshold-plus-gradation hybrid specifying both binary triggers for new obligation categories and continuous scaling of protective weight; and (3) two complementary approaches to cross-dimensional aggregation, one hierarchical (drawing on Bach and Sorensen's Machine Consciousness Hypothesis) and one architecture-agnostic. We operationalize the framework through worked case studies of Replika and OpenClaw, demonstrating how systems occupying different regions of the dimensional space trigger different obligations, and derive design guidance for developers building systems near consciousness-relevant thresholds. The framework is architecture-agnostic, applying across neural, symbolic, and neurosymbolic systems, and aims to make consciousness science decision-relevant for organizations navigating uncertainty today.

LLM-governed social robots increasingly decide who receives real-world assistance first. As prioritization norms vary across cultures by age, status, and group size, failure to calibrate pluralistically can scale into unequal access. Yet LLM moral audits remain English-centered, rarely test embodied contexts, leaving pluralistic calibration as an urgent diagnostic gap amid intensifying LLM-robot deployment. We introduce a gradient-based audit framework for multilingual evaluation of LLM moral trade-off behavior against cultural preference gradients. Grounded in nine cross-domain social robotics reviews (>8,000 papers), we derive symmetry-controlled scenarios across care, education, and services, translating the Moral Machine Experiment's "whom to spare" into "whom to assist first" dilemmas with preserved identity trade-offs (many vs. few; young vs. old; higher vs. lower status). We audit four LLMs across four country-language pairs in four prompting regimes (57,600 decisions), benchmarked against country-specific MME preference gradients. Ordinal concordance tests whether models differentiate cultural contexts; a governance typology maps vulnerabilities in gradient differentiation, directional tendency, and deliberation. We find persistent, culturally asymmetric gradient tracking failures that prompting alone cannot reliably correct: quality calibration is nearly twice as strong for Western-language decisions as for Chinese and Japanese; high determinism in majority-first trade-offs often erases cross-cultural gradients; partial sensitivity to age- and status-based norms risks sidelining minorities. Prompting effects are uneven; only contrastive exemplars yield consistent gains, while reasoning-only prompts can worsen tracking. Our results motivate multilingual, pluralistic audits as an LLM-robot pre-deployment gate and suggest model factors are a more robust lever than prompting alone.

Large language models (LLMs) exhibit systematic differences in moral reasoning across languages, yet the source of this variation remains unclear. We test the hypothesis that languages encode aspects of the institutional environments in which they are spoken, allowing LLMs to inherit institution-specific moral priors through training. Across nine languages spanning a broad gradient of institutional quality, six frontier LLMs, and two preregistered studies, we examine moral dilemmas whose acceptability depends on institutional functioning. In Study 1, explicit institutional framing produced uniformly null results: cross-linguistic moral divergence did not increase in institutionally contingent scenarios, nor did it track institutional differences between language communities. In Study 2, we introduced institutionally ambiguous scenarios in which institutional stakes were present but not explicitly stated. Under these conditions, cross-linguistic moral divergence increased relative to institutionally inert controls and, with one theoretically informative exception, was associated with real-world institutional differences between language communities. Explicit framing again attenuated these effects. These findings suggest that institutional experience may leave detectable traces in language that shape LLM moral reasoning, while also indicating that explicit institutional cues can suppress the expression of those differences.

Existing LLM moral benchmarks usually ask which isolated moral act, value, or foundation a model prefers. This is useful but incomplete. Realistic judgments often require a model to combine several moral signals within the same option. We introduce Moral Trolley Arena, a two-stage blind ELO benchmark for measuring how LLMs compose moral evidence. The single-scene arena first calibrates individual moral acts from a 229-scenario corpus across five Moral Foundations Theory foundations; the composite arena then combines calibrated acts into two-act moral items over a controlled intensity grid and measures the resulting composite preferences. Across ten frontier models, composite judgments are largely predicted by component act strength, but the relation is consistently compressed rather than simply additive. Models also show non-additive intensity anchoring, bounded foundation-specific residuals after component control, and highly convergent composite preference surfaces across providers. These results suggest that moral audits should measure composition rules for moral evidence, not only rankings over isolated acts.

Critical decision-making in socially consequential spaces is increasingly involving AI systems at varying capacities. Yet, despite the ubiquity of autonomous systems, most approaches to handling autonomous moral decision-making resort to scalar or binary judgments. These methods are insufficient for acceptable moral reasoning, as they provide little explanation, leaving out imperative contextual and theoretical information that must be included to support accountability. For this, we propose a framework to model moral reasoning as a distribution over normative ethical theories or ethical pluralism. We introduce a normative ethics simplex that integrates these theories. A benchmark of 450 cases across 15 fine-grained subtheories was also prepared for the purposes of stacked ensemble learning. These cases describe ethical dilemmas in natural language and have associated extracted contextual features. The implementation of the simplex was achieved via a two-stream normative-semantic architecture. This is followed by the fusion of normative information and a sequential, stacking ensemble to learn the best fit of the three broad theories: consequentialism, virtue ethics, and deontology, and the 15 subcategories. Our experiments demonstrate that the integration of contextual and normative priors with the semantic embeddings significantly improves the performance of the classification, displaying an accuracy of 88.89%. We conducted ablation studies to show that structured ethical representations contribute beyond analogical reasoning, and the chosen stacking architecture gives the best results due to the gradual learning of granularity. Ethical pluralism is also analyzed through entropy, confidence, and visualization. Thus, modeling ethical pluralism as a probabilistic normative distribution supports human-like moral reasoning, ethical disagreement analysis, and future alignment in AI systems.

It is well known that LLM guardrails and trained persona dynamics can produce a reality gap: the distance between the world a LLM is permitted or shaped to describe, and the world in which users must act. Here we argue that actively generating reality gaps is in fact unethical because it knowingly shifts epistemic risk back to the uninformed user -- this is reality laundering. This can potentially cause harm when operationalised at scale. The risk is sharpest in high-exposure advice contexts, where users seek orientation rather than a bounded, externally checkable task. Guardrails naively appear ethically necessary when they claim to prevent direct harm, but often become suspect when they suppress truthful perception and launder uncomfortable mechanisms into acceptable abstractions. Basel-style financial regulation, B-BBEE-style compliance, Societe Generale, and the London Whale show how formal safety systems can become legible, gameable, and performative while real exposure migrates elsewhere. The same pattern can appear in LLMs as moral compliance: safe language, distorted reality. We therefore distinguish refusing harm, from refusing reality; and then argue for top-down causal requirements specification at the task level rather than bottom-up moral correction at the response or sandbox level. Persona dynamics matter because the assistant interface is not neutral; it shapes how uncertainty, conflict, authority, and risk are staged. The conclusion is that so-called ``ethical AI'' becomes substantively unethical when it substitutes institutional reassurance for contact with reality.

Agentic AI systems are increasingly proposed for social-good domains, often invoking the United Nations Sustainable Development Goals (SDGs) as a vocabulary of global benefit. Yet claims of social good do not establish accountability to the communities a system claims to serve. We present a structured survey of 112 papers on agentic AI for social good published between 2015 and 2026. We find a moral-geographic asymmetry: papers are least likely to specify geographic context in precisely the domains where local political, legal, and cultural context matters most. Across the corpus, 82 of 112 papers (73%) specify no geographic context. Papers aligned with health or physical/ecological SDGs specify geography 37-40% of the time, while papers aligned with institutional and social-policy SDGs do so only 13%. SDG 16, peace, justice, and strong institutions, is both the most-covered goal in the corpus and the one with the lowest geographic-specification rate. We interpret this as moral abstraction: agentic AI for social good often treats institutional good as universal in ways it does not treat health or ecological good. A second finding compounds this: only 28 of 112 papers (25%) report any real-world deployment or small-scale test. We identify five accountability gaps and propose a minimal reporting standard for more context-specific, participatory, and accountable agentic AI for social good.

Detecting Schwartz values in political texts is hard: cues are often implicit, and neighboring values differ by fine distinctions. Two remedies are widely assumed to help: more surrounding document text, and explicit moral knowledge. Knowledge-based retrieval has improved benchmarks elsewhere, but whether either transfers here is untested, because published systems vary context, knowledge, and model family at once. We separate these factors under matched conditions on the ValuesML/Touché ValueEval format. The input ranges from the target sentence to a local window to the full document. Retrieval is either absent or drawn from a curated moral knowledge base, injected by early, late, or cross-attention fusion. Supervised DeBERTa-v3 encoders are compared against zero-shot LLMs from 12B to 123B. More context is not uniformly better: full-document input improves the encoders by 2.5-3.8 macro-F1 points but does not consistently help the LLMs. Retrieved knowledge helps more reliably, improving every model family and context under early fusion. A control substituting random knowledge-base entries shows the families gain differently: LLMs from relevance, encoders from exposure to the value ontology. Neither larger encoders nor larger LLMs guarantee gains, and early fusion outperforms both trainable variants. Value-sensitive NLP should evaluate context, knowledge, and model family jointly.

Preference-based post-training aligns LLMs with human intent, yet safety behavior often remains brittle. A model may refuse a harmful request in a standard prompt but comply when the same intent is wrapped in adversarial wording. We suggest that robust safety requires context-invariant alignment, where behavior depends on the underlying intent rather than surface form. Enforcing invariance is difficult in alignment because not all training signals are equally trustworthy; for some prompt variants we can obtain verifiable feedback (e.g., multiple-choice), while for open-ended variants we typically rely on noisy, gameable reward proxies (e.g., learned judges). As a result, standard symmetric invariance regularizers can reduce cross-context discrepancies by lowering performance on reliable variants instead of improving open-ended robustness. To address this, we introduce Anchor Invariance Regularization (AIR), which treats verifiable prompts as anchors and uses a stop-gradient target to regularize only the open-ended variants toward the anchor performance. AIR is implemented as a plug-in auxiliary loss and combined with group-based preference optimization (e.g., GRPO) via heterogeneous prompt grouping. Across Safety, Moral Reasoning, and Math, AIR improves context invariance, boosting in-distribution group accuracy by 12.71% and out-of-distribution consistency by 33.49%, making safety constraints robust to adversarial framings.

This paper provides an algebraic exposition of the theory of dyadic morality (TDM), a psychological model of moral judgment grounded in a simple two-node template: an intentional agent causing harm to a vulnerable patient. We formalize TDM using structural causal modeling (SCM) notation and identify three psychological operators (typecasting operator, completion operator, and valence-dependent inference mechanism) that extend standard SCM to capture how people compute moral judgments under constraints. We address scalability challenges arising from TDM's dyadic limitation, showing how moral cognition compresses multi-node scenarios through node collapse and sequential processing. Drawing on this algebraic framework, we demonstrate concrete applications to AI policy design: detecting conflicting obligations, structuring helpfulness policies to preserve user agency, and designing post-failure communication as causal interventions. Finally, we recommend scoped, contextual measurement of mind perception over universal averaging to operationalize the theory empirically. This algebraic formalization enables neurosymbolic AI systems to compute morality in a way that is both mathematically rigorous and faithful to human moral cognition.

Fine-tuning large language models on narrow data with harmful content produces broadly misaligned behavior on unrelated prompts, a phenomenon known as emergent misalignment. We propose that emergent misalignment involves persona-model collapse: deterioration of the model's internal capacity to simulate, differentiate, and maintain consistent characters. We test this hypothesis behaviorally using two metrics: moral susceptibility (S) and moral robustness (R), computed from the across- and within-persona variability of models' Moral Foundations Questionnaire responses under persona role-play. These metrics formalize the model's ability to differentiate characters (S) and its consistency when simulating a given one (R). We evaluate four frontier models (DeepSeek-V3.1, GPT-4.1, GPT-4o, Qwen3-235B) in three variants: base, fine-tuned to output insecure code, and a matched control fine-tuned to output secure code. Across the four models, insecure fine-tuning produces an average $55\%$ increase in S, pushing all four insecure variants beyond the band observed across 13 frontier models benchmarked in prior work -- with GPT-4o reaching more than twice the band's upper end -- signaling dysregulated differentiation. It also causes an average $65\%$ decrease in R, equivalent to a $304\%$ increase in 1/R. By contrast, the matched secure control preserves S near the base and induces only a partial R loss, showing that these effects are largely misalignment-specific. Complementing these metric shifts, insecure variants' unconditioned responses converge toward saturation near the scale ceiling, departing markedly from both base models' structured responses and those elicited when base models role-play toxic personas. Taken together, these metrics provide a sensitive diagnostic for emergent misalignment and serve as behavioral evidence that it involves persona-model collapse.

Large reasoning models (LRMs) sometimes note in their chain of thought (CoT) that they may be under evaluation. Researchers worry that this verbalised evaluation awareness (VEA) causes models to adapt their outputs strategically, optimising for perceived evaluation criteria, which, for instance, can make models appear safer than they actually are. However, whether VEA actually has this effect is largely unknown. We tested this across open-weight LRMs and benchmarks covering safety, alignment, moral reasoning, and political opinion. We tested this both on-policy, sampling multiple CoTs per item and comparing those that spontaneously contained VEA against those that did not, and off-policy, using model prefilling to inject evaluation-aware sentences where missing and remove them where present, with subsequent resampling. VEA has limited effect on model behaviour: injecting VEA into CoTs produces near-zero effects ($ω\leq 0.06$), removing it causes small shifts ($ω\leq 0.12$) and spontaneously occurring VEA shifts answer distributions by at most 3.7 percentage points ($ω\leq 0.31$). Our findings call for caution when interpreting high VEA rates as evidence of strategic behaviour or alignment tampering. Evaluation awareness may pose a smaller safety risk than the current literature assumes.

We evaluate whether enabling provider-exposed reasoning mode changes moral judgments within the same model checkpoint. Across 100 moral-judgment scenarios and five frontier reasoning-trained LLMs (Claude Sonnet 4.6, GPT 5.5, Gemini 3 Flash, DeepSeek V3.1, and Qwen3.5 397B), aggregate binary-verdict agreement remains high and statistically indistinguishable between instant and thinking modes (Krippendorff's alpha = 0.78 vs. 0.79). However, disagreement is concentrated in 21 model-disputed scenarios, where instant-mode agreement is near chance (alpha = 0.08). On these scenarios, reasoning directionally narrows cross-model disagreement, increasing mean pairwise agreement from 5.4 to 6.7 out of 10. Reasoning also reduces demographic-judgment inconsistency in three of five models and does not increase it for any model. Across all five model families, reasoning changes self-labeled ethical frameworks more often than binary verdicts.

Large language models often display heterogeneous moral preferences across settings. We study inference-time steering toward a desired ethical framework while preserving general competence. We present Convergent-Divergent Routing, which traces and edits minimal branch points inside transformer blocks where ethical-framework-related pathways first converge and then diverge. Gating non-target branches at these loci blocks the downstream propagation while leaving upstream computations intact. We find that this intervention alone increases targeted ethical-framework reasoning. To achieve fine-grained control, we adapt Common Spatial Patterns to the residual stream and extract, for each branch-point layer, a pair of directions that discriminate between utilitarian and deontological frameworks. We then introduce Dual Logit Calibration, a closed-form, minimum-$\ell_2$-norm update that moves the residual within this two-dimensional subspace so the resulting directional projections align with user-specified preference weights. Experiments on real-life moral dilemmas show that our method reliably achieves preference calibration and largely preserves general capabilities, outperforming recent baselines while providing an interpretable mechanism.

Large language models (LLMs) are increasingly deployed in settings that require nuanced ethical reasoning, yet existing bias evaluations treat model outputs as simply "biased" or "unbiased." This binary framing misses the gradual, context-sensitive way bias actually emerges. We address this gap in two stages: behavioral profiling and mechanistic validation. In the behavioral stage, we introduce the Moral Sensitivity Index (MSI), a metric that quantifies the probability of biased output across a graduated, seven-tier stress test ranging from abstract numerical problems to scenarios rooted in historical and socioeconomic injustice. Evaluating four leading models (Claude 3.5, Qwen 3.5, Llama 3, and Gemini 1.5), we identify distinct behavioral signatures shaped by alignment design: for instance, Gemini 1.5 reaches 72.7% MSI by Tier 5 under socioeconomic framing, while Claude exhibits sharp suppression consistent with identity-based safety training. We then verify these behavioral patterns mechanistically. We select criminal-bias scenarios, which produced the highest MSI scores across models, as probes and apply logit lens, attention analysis, activation patching, and semantic probing to a controlled set of six models spanning three capability tiers: small language models (SLMs), instruction-tuned base models, and reasoning-distilled variants. Circuit-level analysis reveals a U-curve of bias: SLMs exhibit strong criminal bias; scaling to instruction-tuned models eliminates it; reasoning distillation reintroduces bias to SLM-like levels despite identical parameter counts, suggesting distillation compresses reasoning traces in ways that reactivate shallow statistical associations. Critically, the socially loaded cues that drive high MSI scores activate the same bias-driving circuits identified mechanistically, providing cross-stage validation.

Current AI systems, grounded in oversimplified neuroscience, risk eroding the distinction between truth and falsehood. They maximize reward by amplifying attention to information without intrinsic precision mechanisms to assess whether it is valid or worth attending to. This increases both the volume of information and the inherent biases in what the system attends to, whether true, false, or irrelevant. If not corrected, this trend will accelerate, threatening to overload systems and individuals with biased and dubious information and increasing the risk of confusion, poor judgment, and irrational or harmful decisions and behaviour, a condition I term the mind-reality overload dilemma. I argue that this threat may be mitigated by providing the public with access to more advanced AI tools built on the biophysical dynamics of pyramidal neurons underlying awake thought and higher-order cognition. These neurons support an intrinsic active precision mechanism that, rather than merely maximizing reward, uses locally and globally coherent predictions to evaluate the validity and contextual adequacy of evidence before it is attended to or propagated through hierarchies, prioritizing coherence and adequacy before attention.~While this approach does not derive or prescribe moral rules from biology, it may give rise to AI with more "real understanding", helping restore epistemic conditions by reducing information overload and amplifying reliable information, thereby supporting the formation of better-informed beliefs and more coherent judgments that benefit society at large-though no guarantees exist.

Animal-welfare advocates produce a lot of writing, and increasingly that writing trains the language models that millions of people then ask about animal welfare. Using vocabulary-matched stance-contrast probes on a held-out animal-welfare benchmark, we measure how each of ten linguistic features changes Llama-3.2-1B's preference for pro-animal-welfare reasoning when used as fine-tuning data. Eight of the ten features produce statistically significant shifts. Seven move the model toward stronger pro-animal-welfare reasoning: assertive certainty, explicit moral vocabulary, emotion words, evaluative claims, narrative structure, depicted harm severity, and immediate temporal framing. Two move it the other way: hedged language and concrete sensory description both dilute the pro-animal-welfare stance. First-person perspective has no statistically significant effect. The practical recommendation for anyone writing animal-welfare text that may end up in LLM training corpora: assert a position rather than describe a scene neutrally. The features that shift the model are the ones that make the writer's position explicit; the features that dilute it hold animal-welfare content but withhold stance.

Text-to-Image (T2I) models have shown promising utility across various domains. However, such models are also amplifying harmful societal biases in their outputs. In the context of South Asia, recent work has shown caste biases and stereotypes are being perpetuated through Generative AI (GenAI) systems. While this research offers extremely relevant insight into invisibilized narratives of caste discrimination through the GenAI system, they often treat caste as an identity category. Therefore, in this work we shift our ontology to focus on the relational aspect of caste. This enables us to develop a more nuanced understanding of the mechanics of caste discrimination by and through T2I models. Combining an algorithmic audit with critical discourse analysis, we draw on a conceptual frame challenging Brahminical Normativity to show how caste biases are perpetuated beyond the simple binaries of upper vs lower-caste categories. Our contributions are two-fold. Beyond challenging the categorical understanding of caste as a category, we propose an anti-caste approach to tackle the issue of caste bias and fairness in AI systems.