Ghost in the Context: Measuring Policy-Carriage Failures in Decision-Time Assembly
Authors: Igor Santos-Grueiro
Summary
arXiv:2605. 12535v1 Announce Type: new Abstract: LM agents do not act on raw interaction history; they act on a bounded decision state assembled by truncation, summarization, reordering, and rewriting.
Relevance
Read next because Ghost in the Context: Measuring Policy-Carriage Failures in Decision-Time Assembly overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: text, rect, strong, directive, under, eval, prefix, prompt. Source: arxiv cs.CR (Cryptography and Security).
Threat model
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses failure, failures.
Abstract
arXiv:2605.12535v1 Announce Type: new Abstract: LM agents do not act on raw interaction history; they act on a bounded decision state assembled by truncation, summarization, reordering, and rewriting. If directive-bearing state is dropped, weakened, or rebound during that step, an agent can cross a policy boundary without prompt override, model changes, or persistent-memory compromise. We study this failure mode over local Llama 3.1 8B, Qwen 2.5 7B, and Mistral 7B using judged exact constraint respect and direct audits of assembled-state visibility. We evaluate SafeContext, a control layer that pins control state, reuses retained control prefixes, and optionally injects reminders under pressure while keeping model weights fixed. Unmitigated risk is systematic, but absolute exact respect remains low. Against truncation, SafeContext yields small gains; against a strong structured-compaction policy, most aggregate lift disappears, leaving residual benefit mainly in overflow eviction and selected aliasing slices. Replay-only does not explain the effect. A larger-model extension on Qwen 14B and Llama 70B shows the same failure object under larger models, although sign and magnitude remain policy-conditional. Decision-time context assembly is therefore a measurable part of the control path that can be partially hardened.