ainewsblitz.com

Breaking

Google DeepMind pilots double-blind evaluations of Gemini 2.5 Flash Lite

  • Foundation Models
  • Security
  • Policy & Regulation

Google DeepMind is piloting what it describes as the world’s first double-blind evaluation of a proprietary, frontier-class AI model, running Gemini 2.5 Flash Lite against confidential safety benchmarks inside a cryptographically attested enclave so that neither test prompts nor model weights are revealed to the other party. The company says the setup is meant to keep external safety and performance tests private, robust, and harder to game through benchmark contamination.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.

$20
Read this article
$29/month
Unlimited — all 27,504 articles, the full archive, and comprehension quizzes
Save 72%
$98/year
≈ $8.17/month
Unlimited, billed once a year