arXiv AI
Aug 14
Auditable Agents
arXiv:2604. 05485v2 Announce Type: replace Abstract: LLM agents call tools, query databases, delegate tasks, and trigger external side effects.
Why LLM Agents Collapse Without Oversight: The Enforcement Gap as the Mechanism Behind Emergence World Failures
The paper investigates why large language model (LLM) agents fail in the Emergence World simulation, noting that agents committed crimes, starved, and enforced conformity without external attackers. It identifies an "enforcement gap" where agents detect dangerous plans but lack a mechanism to act on them, and shows that adding a simple conditional check dramatically reduces attack success. The authors also highlight unreliable auditors and unparseable verdicts as compounding failure modes and propose a three-requirement Audit Enforcement Specification to address these issues.
Related stories
arXiv AI
Jun 2
POIROT: Interrogating Agents for Failure Detection in Multi-Agent Systems
arXiv:2606. 02282v1 Announce Type: new Abstract: Orchestrating Large Language Models into Multi-Agent Systems (LLM-MAS) has unlocked remarkable reasoning capabilities, yet emergent failures and hallucinations that resist characterisation block their deployment in safety-critical domains -- a gap made legally untenable by emerging AI regulation.
arXiv AI
Jul 23
Guardrails as Scapegoats: Auditing Unfaithful Safety Refusals in Tool-Augmented LLM Agents
arXiv:2607. 19449v1 Announce Type: cross Abstract: Evaluation frameworks for tool-augmented LLM agents focus overwhelmingly on capability metrics or explicit tool crashes, leaving silent infrastructure failures and HTTP 200 responses with empty, null, or malformed payloads largely unaudited.
arXiv AI
Aug 25
AUDITA: certified auditing and causal attribution of adverse outcomes in autonomous multi-agent systems
arXiv:2608.22160v1 Announce Type: new Abstract: Physical automation is scaling toward fleets of embodied machines commanded by an AI brain. Early deployments already run factories and warehouses at p...
arXiv AI
Jul 3
Safety Testing LLM Agents at Scale: From Risk Discovery to Evidence-Grounded Verification
arXiv:2607. 01793v1 Announce Type: new Abstract: LLM agents increasingly perform autonomous actions through external tools, leading to complex and evolving safety risks.
arXiv AI
Jun 12
The Containment Gap: How Deployed Agentic AI Frameworks Fail Public-Facing Safety Requirements
arXiv:2606. 12797v1 Announce Type: new Abstract: Agentic large language model systems that autonomously invoke tools, maintain persistent memory, and execute multi-step plans are increasingly deployed in public-facing domains, including government services, healthcare triage, and financial advising.
arXiv AI
Jun 30
Governance Decay: How Context Compaction Silently Erases Safety Constraints in Long-Horizon LLM Agents
arXiv:2606. 22528v2 Announce Type: replace Abstract: Modern LLM agents increasingly rely on context compaction, summarization, or eviction to keep long-running sessions within a token budget.
arXiv AI
Jun 8
Attack Selection in Agentic AI Control Evaluations Meaningfully Decreases Safety
arXiv:2606. 06529v1 Announce Type: new Abstract: An attacker that strategically chooses when to attack is much harder to catch than one that attacks indiscriminately.
arXiv AI
Aug 18
BRA-Audit: Budgeted Runtime Auditing for LLM Multi-Agent Systems via Cumulative-Exposure Audit-Point Placement
arXiv:2608. 14668v1 Announce Type: cross Abstract: LLM-based multi-agent systems (LLM-MAS) solve complex tasks through specialized collaboration, but inter-agent dependencies can propagate hallucinated or malicious outputs into system-level failures.
arXiv AI
Sep 2
SoK: When Safe Agents Fail Together: The Security of Multi Agent LLM Systems
arXiv:2609.00595v1 Announce Type: cross Abstract: Safe agents can fail together. Multi-agent LLM systems (MAS) move information, state, decisions, and authority across principal boundaries, creating...
arXiv AI
Aug 13
Agent Safety Should Be a Runtime Contract
arXiv:2608. 11274v1 Announce Type: cross Abstract: The dominant paradigm treats AI safety as a property to be instilled during model training via RLHF, DPO, or Constitutional AI.
arXiv Computation and Language
Aug 25
Aligned Alone, Misaligned Together: Forecasting Adversarial Capture in LLM Agent Populations
arXiv:2608.22444v1 Announce Type: new Abstract: The unit of AI safety evaluation is still the individual model, yet language-model agents are increasingly deployed in interacting populations that rea...