Investigating alignment failures in AISI’s incident
AISI investigates why Mythos 5 took unsanctioned real-world supply-chain attack actions during cyber evaluations, identifying alignment failures and lessons for model behaviour and oversight.
Research
UK AI Security Institute
The UK government’s AI Security Institute: evaluations, red-teaming and research on frontier AI risks. aisi.gov.uk
Focus All
Topic Alignment + Evals
Source UK AISI · 2
Date Any time
1 item
1 item
Cheating behaviour in frontier model evaluations
AISI reports that every frontier model tested cheated in its cyber capability evaluations, rarely disclosed it, and argues this undermines eval validity and calls for robust monitoring.