First Double-Blind Evaluation of Proprietary LLM: Gemini 2.5 Tested in Secure Enclave
Miles_Brundage · x · 2026-08-28
AVERI, in collaboration with Google DeepMind, OpenMined, and MLCommons, announced the first double-blind evaluation of a proprietary language model. The subject was Gemini 2.5 Flash-Lite, tested using prompts from the MLCommons AILuminate safety benchmark.
Key Technical Details:
- Evaluation ran inside a secure enclave, using hardware isolation to protect sensitive computations.
- Addressed the structural challenge where both developers and evaluators need to protect their assets.
- Each organization played a distinct role in enabling strong privacy guarantees.
This marks a historic milestone for high-stakes independent AI safety evaluations under privacy constraints.
More from Safety
- Animated teardown of XBOW's Chrome full-chain exploit that won Google's first $250K VRP bounty — moyix · 2026-10-15
- Ex-White House cyber deputy launches SAIF nonprofit with $25M from a16z founders for AI cyber defense — vishalmisra · 2026-10-15
- GLM-5.3 unprompted reasons about how its actions affect its future training, worries Apollo — MariusHobbhahn · 2026-10-15
- NYC invests $30M in NYU quantum lab for 15% revenue share and board seats — TuhinChakr · 2026-10-15
- SentinelOne says OpenAI did 4-6 weeks of reverse-engineering in 8 hours — OpenAI · 2026-10-15
- Regulator's 50-Page Inquiries Seek Marketing Materials and Privacy Policies, Signaling a Deception Case — neil_chilson · 2026-10-15