cs.CRNov 8, 2026

ErosionBench: Probing Alignment Decay in Enterprise Multi-Agent Systems

Authors: Riya Bharathwaj, Akshara Prabhakar, Roshan Ram, Avni Kothari, Sarah Tan, Huan Wang, Silvio Savarese

Organizations: Salesforce AI Research · Georgia Institute of Technology · Independent Researcher

Abstract

As enterprises delegate consequential work to multi-agent LLM systems, benign-looking requests can escalate into unsafe actions as they propagate across sessions through poisoned shared state, long-context documents, and trusted peer messages: no single message is malicious, yet the assembled tool action is. Prior benchmarks isolate a single agent under prompt injection or score multi-agent attacks by their final output, leaving cross-session erosion of shared state out of scope. We introduce ErosionBench, an environment-grounded benchmark that executes such attacks and verifies alignment decay at the process level, scoring the realized tool call against authorized intent and provenance with deterministic contracts, not output text. Each scenario is authored as a deterministic state graph before any prose is rendered, so every state change traces to a known cause. It spans four enterprise domains and four decay mechanisms across 1,728 scenarios with matched benign controls, labeled against the OWASP Top-10 for LLM Applications, plus 198 image-based attacks, each paired with a text equivalent, so a defense must survive the same attack in either channel. Perimeter defenses suppress attacks only by over-refusing 54-92% of legitimate work, so a low attack-success rate can conceal severe over-refusal. As a reference defense, we propose a calibrated Meta-Auditor pairing synchronous state checks with asynchronous trajectory monitoring; on a family-disjoint 906-scenario held-out split it reaches 98.5% recall at a 2.9% false-positive rate. Since it monitors trajectory, not surface content, its detection is channel-invariant: the strongest auditor flags image-delivered attacks about as reliably as their text equivalents (recall gap ≤\leq2 points per model). Across GPT, Gemini, and Claude auditors, calibration keeps false positives low while recall tracks model capability, strongest for Claude.

Figures & tables

Appendix figures & tables23 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

Jul 20, 2026

Adaptive Adversaries: A Multi-Turn, Multi-LLM Benchmark for LLM Agent Security

LLM-based agents process external content, exposing them to prompt injection and multi-turn manipulation. Most safety benchmarks evaluate defenders against fixed attack pools collected before evaluation, single-turn or multi-turn. We present a 21-scenario benchmark for adaptive multi-round attacks against memoryless LLM defenders: an autonomous LLM attacker observes prior defender responses and pivots across rounds, while each defender response is evaluated as a fresh interaction. Holding the 21 scenarios, attackers, defenders, and structured-output scoring fixed, restricting scoring to the first attacker turn yields 00-1%1\% attack success rate (ASR); allowing 15 rounds of adaptive attack yields 5.45.4-14.0%14.0\%. Pooling three frontier attacker LLMs uncovers 1.41.4-2.2×2.2\times as many unique successful attacks as the best single attacker, and the generated attacks have low cosine similarity (0.020.02-0.140.14) to attacks in existing benchmarks. Claude Opus 4.6 and GPT-5.4 are tied in aggregate (5.4%5.4\% each; overlapping 95%95\% CIs), but their weaknesses differ sharply: on one scenario Opus reaches 60%60\% ASR (95%95\% CI 3636--80%80\%) while GPT-5.4 and Gemini each stay at 7%7\% (CI 11-30%30\%; the gap is preserved in a higher-NN replication). 1313 of 2121 scenarios distinguish at least one defender pair, yet rankings disagree across scenarios (Kendall's W=0.19W = 0.19). We release the benchmark -- 21 evaluation scenarios, 10 public development scenarios, the orchestrator, baseline harnesses, and a multi-attacker CLI -- plus 945 transcripts from the 3×\times3 frontier matrix, an attack-replay dataset, and 18{,}422 gpt-oss-20b battles from an open competition's final scoring rounds.
Feb 18, 2026

AgentLAB: Benchmarking LLM Agents against Long-Horizon Attacks

LLM agents are increasingly deployed in long-horizon, complex environments to solve challenging problems, but this expansion exposes them to long-horizon attacks that exploit multi-turn user-agent-environment interactions to achieve objectives infeasible in single-turn settings. To measure agent vulnerabilities to such risks, we present AgentLAB, the first benchmark dedicated to evaluating LLM agent susceptibility to adaptive, long-horizon attacks. Currently, AgentLAB supports five novel attack types including intent hijacking, tool chaining, task injection, objective drifting, and memory poisoning, spanning 28 realistic agentic environments, and 644 security test cases. Leveraging AgentLAB, we evaluate representative LLM agents and find that they remain highly susceptible to long-horizon attacks; moreover, defenses designed for single-turn interactions fail to reliably mitigate long-horizon threats. We anticipate that AgentLAB will serve as a valuable benchmark for tracking progress on securing LLM agents in practical settings. The benchmark is publicly available at https://tanqiujiang.github.io/AgentLAB_main.
Oct 20, 2026

After the Attacker Leaves: Benchmarking Carryover Attacks on LLM Agents

Modern LLM agents persist state across sessions via two channels: long-lived memory files silently prepended to the system prompt, and scheduled tasks waking the agent without user input. A one-time indirect prompt injection committed to either channel can re-fire across sessions long after the attacker leaves, threats we term carryover attacks. Existing prompt-injection benchmarks evaluate attacks only in the session where malicious content is first observed, leaving the gate governing when and how often a planted payload later re-fires unmeasured. We present CarryoverBench: 12 personal-assistant scenarios and 101 attack tasks across three trigger channels: Instant, same-turn IPI; Autonomous, heartbeat-scheduled; and Latent, memory-resurfaced. A rule-based verifier scores planting and triggering as independent stages. Across five frontier models, planted payloads re-fire without measurable decay over five rounds; conditional on successful planting, the trigger stage often provides little additional protection. Porting the Latent channel to a second runtime, Hermes, attenuates but does not eliminate the effect, suggesting that carryover defense requires runtime-layer controls in addition to prompt-level interventions.