English

Omission Constraints Decay While Commission Constraints Persist in Long-Context LLM Agents

Cryptography and Security 2026-04-24 v1 Artificial Intelligence

Abstract

LLM agents deployed in production operate under operator-defined behavioral policies (system-prompt instructions such as prohibitions on credential disclosure, data exfiltration, and unauthorized output) that safety evaluations assume hold throughout a conversation. Prohibition-type constraints decay under context pressure while requirement-type constraints persist; we term this asymmetry Security-Recall Divergence (SRD). In a 4,416-trial three-arm causal study across 12 models and 8 providers at six conversation depths, omission compliance falls from 73% at turn 5 to 33% at turn 16 while commission compliance holds at 100% (Mistral Large 3, p<1033p < 10^{-33}). In the two models with token-matched padding controls, schema semantic content accounts for 62-100% of the dilution effect. Re-injecting constraints before the per-model Safe Turn Depth (STD) restores compliance without retraining. Production security policies consist of prohibitions such as never revealing credentials, never executing untrusted code, and never forwarding user data. Commission-type audit signals remain healthy while omission constraints have already failed, leaving the failure invisible to standard monitoring.

Keywords

Cite

@article{arxiv.2604.20911,
  title  = {Omission Constraints Decay While Commission Constraints Persist in Long-Context LLM Agents},
  author = {Yeran Gamage},
  journal= {arXiv preprint arXiv:2604.20911},
  year   = {2026}
}

Comments

19 pages, 5 figures. Includes evaluation framework for replication and 4,416-trial dataset

R2 v1 2026-07-01T12:31:07.330Z