AI models can provably hide skills from safety tests
Why it matters
Proof-of-concept (NeurIPS 2026): an encrypted network hidden in a model's weights runs only on a secret trigger and stays provably hard to elicit. After fine-tuning it scored 50.9% vs 50.6% for a random control; short 8-bit triggers could still be brute-forced.
Read the source: arXiv (opens in a new tab)
All hits from 13 Oct 2026More SecurityGet the daily hits by email