Dario's "Pace the Frontier" essay lands as DeepMind pilots first double-blind AI evaluations

sebkrier · x · 2026-09-14

Anthropic CEO Dario Amodei published a new essay, We Must Pace the Frontier, arguing the AI industry should slow down, with a three-part plan. Anthropic is unilaterally committing to the first step: granting third-party evaluators permanent, employee-level access to its systems to verify safety measures, report incidents, and assess alignment during training.

The same day, Google DeepMind announced the world's first double-blind evaluation of a frontier model: partnering with the Singapore AI Safety Institute, OpenMined, AVERI and MLCommons, it tests Gemini Flash Lite against confidential benchmarks inside a cryptographically sealed environment, preventing benchmark contamination.

iamtrask's take: embedded third-party evaluators are "a 1960s solution to a 2026 problem" — the real answer is structured transparency, citing Drexler's paradigm.

Three frontier labs converging on credible evaluation and restraint of frontier models on the same day marks a key moment for AI safety governance.

Related event: Anthropic CEO Calls for Slowing Frontier AI, Opens Third-Party Evaluation First(198 posts)→

Original post →

More from Companies & People

Companies & People channel →