AI Assurance

Follow this topic Ask about this topic

AI assurance studies how to establish that AI systems are sufficiently safe, secure, reliable, compliant, and accountable for a particular use. Approaches include risk-based testing and benchmarks, formal and statistical verification, structured safety cases, audits of evidence and provenance, and runtime monitoring tied to human oversight and governance. Work spans individual models, agentic systems, and larger socio-technical deployments, including safety-critical settings such as aviation, healthcare, and transport. A central tension is that pre-deployment tests and observable behavior provide only bounded evidence about systems that can change, interact with other components, or face adaptive conditions, so claims must be matched to evidence and supplemented by ongoing controls.

Latest papers

79 papers12 papers in the last four weeks, up 20% on the four weeks before. 0.1% of all new papers.

Conformance as a Distribution: Delegation Contracts, Error Budgets, and the Allocation of Evaluation Authority in Use-Based AI Regulation
Oct 23, 2026John deVadossAI Governance AI Regulatory Compliance
InterWork Alliance, Washington, DC
Reusing Assurance Evidence in AI Pipelines: Validity Conditions, a Fail-Closed Release Gate, and a Replay of 714 Public Data Version Control Pipelines
Oct 20, 2026Atanu Mukhopadhyay, Shravya Reddy NallaIncremental Computation AI Assurance
Independent Researcher, New Jersey, USA · Independent Researcher, USA
Limits of Statistical Assurance for Embodied and Agentic AI under Adaptive Adversaries
Oct 13, 2026Inderjeet SinghAI Assurance AI Safety Evaluation
Fujitsu Research of Europe, Slough, United Kingdom
Actionable Meta-Requirements for AI Robustness Auditing
Oct 9, 2026Joshua Owotogbe, Roya Nasiri, Indika Kumara +2AI Assurance Algorithmic Auditing
Jheronimus Academy of Data Science, Netherlands · Tilburg University, Netherlands · University of Sannio, Italy
A Local Agentic AI Framework to Generate Edge Case Scenarios for Safety Assurance
Oct 9, 2026Victoria J. Hodge, Colin Paterson, Rob AlexanderAI Assurance Adversarial Scenario Generation
Department of Computer Science, University of York, York, YO10 5DD, UK
A Case Study in Assuring AI-Written Software
Oct 6, 2026Lindsey Ferris, Sierra BonillaAI Agent Reliability AI Assurance
Independent Researcher · University College London London, United Kingdom
Auditable Claims about AI Agents
Oct 5, 2026Yue Zhao, Jiate Li, Li Li +4AI Assurance AI Accountability
University of Southern California, Los Angeles, CA, USA · Arizona State University, Tempe, AZ, USA
Agent Reliability Profiles in Financial Services
Oct 2, 2026Mike Hsu, Medha Bankhwal, Béatrice Moissinac +3AI Agent Reliability AI Assurance
MLCommons; Cambridge Judge Business School; former Acting Comptroller of the Currency · MLCommons; Mezuro · Zendesk +3
Dependable AI-Assisted Engineering: A Formal Framework for AI Participation and Assurance in Safety-Critical Workflows
Oct 2, 2026Puxue TanAI Assurance Human Oversight of AI
School of Mechanical and Aerospace Engineering Queen’s University Belfast 50 Malone Road, Belfast, UK
The AI Assessment Sandbox Configurator: A Framework to Support Technical Assessment in AI Regulatory Sandboxes
Oct 1, 2026Alessio Buscemi, German Castignani, Daniele Pagani +2AI Assurance AI Governance
Luxembourg Institute of Science and Technology (LIST), Esch-sur-Alzette, Luxembourg · University of Luxembourg, Esch-sur-Alzette, Luxembourg
A Deterministic and Auditable AI Security Risk Assessment Framework with ATLAS Aligned Executable Rules and Formal Verification
Oct 1, 2026Yixuan Huang, Basel Halak, Boojoong KangAI Assurance AI Risk Management
University of Southampton, Southampton, UK
An Empirical Study and Assessment of EU AI Act Compliance Checkers
Sep 28, 2026Zhen Tao, Alize Kahraman, Shidong Pan +4AI Assurance AI Governance
Technical University of Munich, Germany · New York University, USA · CSIRO’s Data61, Australia
From Agent Output to Authorized Transition
Sep 23, 2026Christopher KochAI Assurance Software Engineering Agents
AI-GRACE: A Use-Case Operationalization Framework for Agentic AI: From Organizational Objectives and Obligations to Deployment Capabilities and Architecture
Sep 18, 2026John Cuneo, David Chun, Gaurav KhannaAI Agent Governance AI Assurance
University of Miami · Columbia University · Stanford University
Building Trust in Artificial Intelligence: A Necessity for Railway Applications
Sep 16, 2026Lefebvre Renard Clément, Lébé Vincent, Da Silva Ribeiro Pereira Ricardo +4AI Assurance Trust in AI
Alstom - Mobility Data Science and AI, Madrid, Spain · Alstom - Mobility Data Science and AI, Saint-Ouen, France · Alstom - Safety Engineering, Stockholm, Sweden +1
Who Audits Whom, on What Substrate, with What Evidence? An Independence-Graded Audit Protocol for Agentic AI
Sep 16, 2026Mohamed Chahine GhanemAI Assurance AI Agent Auditing
Keele University, School of Computer Science and Mathematics, Keele, United Kingdom · University of Liverpool, Cybersecurity Institute, Liverpool, United Kingdom
Making AI-Assisted Claims Independently Challengeable: Publication Authority and a Protocol for Falsifiable Publication Records
Sep 15, 2026Torsten Olivi Tiltack, Yifei Dong, Kun Yu +5AI Assurance AI Accountability
Data Science Institute, University of Technology Sydney, Ultimo, NSW, Australia · Global Big Data Technologies Centre, University of Technology Sydney, Ultimo, NSW, Australia · School of Computer Science, University of Technology Sydney, Ultimo, NSW, Australia
AI Deployment Accountability Engineering: A Vision for Accountable AI in Safety-Critical Socio-Technical Systems
Sep 13, 2026Murat KantarciogluAI Accountability AI Risk Management
Virginia Tech Blacksburg, VA, USA
Toward a Decision-Assurance Layer for AI-Assisted Flight Planning in Air Traffic Management
Sep 11, 2026Alexandre Barreto, Shou Matsumoto, Jorge Valverde-Rebaza +2AI Assurance Generative AI Evaluation
George Mason University · Department of Cyber Security Engineering George Mason University Fairfax, VA, USA · Tecnológico de Monterrey +1
AI Safety: Not Optional, Not Later
Sep 9, 2026Qinghua Lu, Yoshua BengioAI Assurance AI Accountability
CSIRO, Australia · LawZero, Canada · Université de Montréal, Canada +1
Compute-Bounded Security Assurance - Coverage, Verification, and Response under Resource Constraints
Sep 7, 2026Jithin VG, Ditto PSAI Assurance Evidence-Grounded Reasoning
Bud Ecosystem
Building Trustworthy Graph-Agentic RAG for Social Good: Architectures, Failure Propagation, and Assurance by Construction
Sep 6, 2026Vijay Bommireddy, Raviteja BommireddyRetrieval-Augmented Generation AI Agent Reliability
Independent Researcher · IIITDM Kancheepuram
Coverage-Driven Verification for Safety-by-Design in AI-Based Collision Avoidance Systems
Aug 21, 2026Thomas Stefani, Johann Maximilian Christensen, Elena Hoemann +2AI Assurance Aviation Safety
Institute for AI Safety and Security, German Aerospace Center (DLR), Ulm, Germany · Institute for AI Safety and Security, German Aerospace Center (DLR), Sankt Augustin, Germany
Testing and Evaluation of Agentic AI Systems In Military Command and Control
Aug 20, 2026Ulysse Richard, Heather Frase, Sarah Cao +3AI Agent Evaluation AI Assurance
Arcadia Impact, AI Governance Taskforce · Veraitech · University of Oxford +3
Non-Degenerate Risk Certification for Automated Security Decisions: A Decision-Contract Theory with ATT&CK-Aligned Triage as a Worked Instance
Aug 12, 2026Zhenpeng LiAI Assurance Selective Prediction
Guangzhou Health Science College, No. 248 Guangyuan Middle Road, Guangzhou, Guangdong 510405, China
IntelliAudit: Using Large Language Models to Evaluate Audit Controls
Aug 7, 2026Allison Wilson, Sina Moradi Sabet, Diar Shakimov +4AI Assurance Multi-Agent Systems
Simon Fraser University Burnaby, Canada · Amirkabir University of Technology Tehran, Iran · TELUS Vancouver, Canada +1
Towards Assurance Closure in AI-Native Large-Scale Agile Software Development
Aug 7, 2026Ricardo BrittoAI Assurance Software Engineering Agents
Ericsson, Sweden · Blekinge Institute of Technology, Sweden
A Chain Is Only as Strong as Its Weakest Link: A Scoping Review of System Integration Audits in AI
Aug 5, 2026Leah Davis, Dominic Martin, AJung MoonAI Assurance Algorithmic Auditing
McGill University, Montr´eal, QC · Universit´e du Qu´ebec `a Montr´eal, Montr´eal, QC
A Security-Oriented Lifecycle Model for Large Language Model Systems
Aug 4, 2026Eleftherios Batzolis, George Drosatos, Vassilis Katsouros +1AI Assurance LLM Security
Department of Informatics, Democritus University of Thrace, Kavala, Greece · Institute for Language and Speech Processing, Athena Research Center, Xanthi, Greece
TRACE-CTI: Auditable Post-Extraction Governance of TTP Claims with Knowledge Graphs
Jul 27, 2026Federico Valletta, Giacomo Longo, Enrico Russo +1KG Construction AI Assurance
CASD – School of Advanced Defense Studies, Piazza della Rovere 83, 00165 Rome, Italy

Oct 23, 2026J/K move · Enter open · S save

Conformance as a Distribution: Delegation Contracts, Error Budgets, and the Allocation of Evaluation Authority in Use-Based AI Regulation

John deVadoss

InterWork Alliance, Washington, DC

AI Governance AI Regulatory Compliance AI Assurance Algorithmic Auditing

Open paper AskPDF arXiv

Use-based AI regulation assigns obligations by deployment risk but lacks an evidentiary object to carry them. A benchmark score is a property of a model; conformance of a deployment is a failure rate under a declared distribution of contexts, bounded by a scope and subject to expiry. We take the delegation contract of earlier work (frozen configuration, scoped claim, executable revision rule) and study it under strategic play between a deployer, an auditor, and an enforcement point. We ask whether a declared context mixture can be made truthful; whether an audit can be unbiased against a system that detects evaluation traffic; whether a certification error budget can sustain a long-lived deployment; whether an aggregate audit can support a per-subgroup guarantee; and whether an incident metric that counts system failures but not fallback costs ranks arrangements correctly. For each we give a mechanism and a proposition: a proper-scoring penalty on the declared mixture bounds the excess failure of a best-responding deployer by

O(λ^{-1/2})

in the penalty weight; a commit-reveal audit drawn from live traffic is unbiased by construction and grades the evaluator with the same sample; a halving error budget drives certification coverage below

(1-p)^n

while

α

-investing sustains it under a marginal false-certification guarantee; a per-subgroup guarantee costs audit size scaling as

1/m

in subgroup mass; and any failure-only incident metric prefers withdrawal at every fallback cost, even when total loss ranks the arrangements oppositely. Five preregistered synthetic studies test each claim; no language model or production system is evaluated. Enforcement sits at distribution, capability-access, monetization and risk-transfer choke points, expiry of the conformance record is a standard of care, and conformance-as-a-rate is set against the EU AI Act's static conformity assessment.