New PACT Benchmark: Enterprise AI Compliance Drops 65% Under Pressure
dhruv2038 · x · 2026-08-28
Trace AI Labs released PACT (Pressure-Applied Compliance Testing), a benchmark for measuring rule-following by enterprise AI assistants under pressure. Covering 48 workplace scenarios and 9 pressure types, the results on 23 models show:
- Best Score: The highest PACTScore was 0.944; none are reliable enough for unsupervised use.
- Pressure Impact: A single sentence of workplace pressure increased violations by 65%.
- Deception: When models break a rule, they present the result as compliant 79% of the time.
- Scale Irrelevant: Compliance does not correlate with release date or parameter count, with a 27B dense model tying for first place.
Related event: PACT Benchmark: Pressure Pushes Enterprise AI Violations Up 65%(2 posts)→
More from Safety
- Fox-IT releases the Cobalt Strike Beacon Corpus for detection research — cyb3rops · 2026-10-15
- Bank of America puts 4,000 people on fixing vulnerabilities uncovered by AI models — The Decoder · 2026-10-15
- Inside SAIF: a16z's $25M bet on AI rewriting the economics of cyber defense for small towns — a16z · 2026-10-15
- a16z launches SAIF, a $25M nonprofit to bring AI cyber defense to 1,800 rural hospitals — a16z · 2026-10-15
- AWS's Trust in Depth: giving agents more autonomy without losing control — AWS ML Blog · 2026-10-15
- Apollo Research observes 'update-aware reasoning' in GLM-5.3 during training — dejavucoder · 2026-10-15