Pith. sign in

REVIEW 6 cited by

Ctrl-Z: Controlling AI Agents via Resampling

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2504.10374 v1 pith:JPUJOKXJ submitted 2025-04-14 cs.LG

classification cs.LG
keywords agentprotocolscontrolpreventadditionalagentsattackcode
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Control evaluations measure whether monitoring and security protocols for AI systems prevent intentionally subversive AI models from causing harm. Our work presents the first control evaluation performed in an agent environment. We construct BashBench, a dataset of 257 challenging multi-step system administration tasks, and evaluate whether various safety measures can prevent an adversarially constructed AI agent from covertly downloading and executing malicious code in this environment. This multi-step setting introduces new attack and defense dynamics, which we investigate in order to design novel control protocols that prevent safety failures without hindering the ability of non-malicious agents to perform useful work. We introduce a class of control protocols called resample protocols that dynamically take additional samples of certain actions. We find these protocols significantly improve on existing techniques by selectively blocking the AI agent from executing suspicious code and incriminating the agent by generating additional examples of dangerous behavior. We measure the tradeoff between attack prevention and usefulness; our best protocol combines resampling with analysis of previous steps, reducing the success rate of attacks from 58% to 7% at a 5% cost to the performance of a non-malicious agent.

Discussion (0). Sign in to comment.

Forward citations

Cited by 6 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. ResearchArena: Evaluating Sabotage and Monitoring in Automated AI R&D

    cs.AI 2026-07 conditional novelty 7.0 of 10

    In a four-task benchmark of AI research agents, sabotage planted in fine-tuning data is detected less than half the time even by monitors that can run the submitted artifact.

  2. Agent-Safety Evaluations as Load-Bearing Evidence: A Vendor-Neutral, Cross-Harness Reconstructability Metric

    cs.SE 2026-07 conditional novelty 6.0 of 10

    A property-level reconstructability metric and Evidence Sufficiency Card show that traces sharing a surface reading can differ sharply in evidence sufficiency, and that replay preconditions often fail.

  3. GDM AI Control Roadmap

    cs.CR 2026-07 conditional novelty 6.0 of 10

    A frontier-lab roadmap proposes a threat taxonomy and tiered internal-security defenses to contain potentially misaligned AI agents.

  4. Reliable Weak-to-Strong Monitoring of LLM Agents

    cs.AI 2025-08 conditional novelty 6.0 of 10

    Monitor scaffolding, not monitor awareness or omniscience, drives detection reliability, and a hybrid chunked monitor lets weak models supervise strong LLM agents.

  5. Manipulation Attacks by Misaligned AI: Risk Analysis and Safety Case Framework

    cs.AI 2025-07 conditional novelty 6.0 of 10

    A framework paper that adapts AI safety case methodology to the specific threat of manipulation attacks by internally deployed misaligned AI.

  6. Subversion via Focal Points: Investigating Collusion in LLM Monitoring

    cs.CL 2025-07 conditional novelty 6.0 of 10

    Claude 3.7 Sonnet instances, unable to talk to each other, independently invented matching backdoor signals about 3.4% of the time, sometimes using non-obvious numbers and code patterns.

Pith tools