Breaking

Demis Hassabis Endorses Dario Amodei's Call to Pace Frontier AI

September 13, 2026 at 04:46 EDT

Policy & Regulation
Foundation Models
AI Agents

Demis Hassabis, co-founder and CEO of Google DeepMind, has endorsed the direction of Anthropic CEO Dario Amodei's proposal to pace frontier AI development, while saying its details still need to be worked out. The alignment puts two leading lab heads behind a safety-first approach that would give evaluation and safeguards more time to catch up with capability gains.

September 12, 2026 · Google DeepMind

Hassabis backs a safer pace for frontier AI

Demis Hassabis says Dario Amodei's direction is right, while the details remain unresolved. The shared focus is time for safety work, outside evaluation, and coordination.

The proposal in numbers

3
steps in Amodei's plan

30
days before release

6–12
months in the warning

How wide is the warning window?

Amodei cited a 6–12-month window for a more capable misaligned swarm scenario.

6 months
lower end

12 months
upper end

The upper bound is twice the lower bound.

Three steps to pace the frontier

Embedded
evaluators

→

Democratic
coordination

→

Global
coordination

The first step is a company commitment; the next two require wider coordination.

Where the consensus stops

Shared ground

Third-party testing can add accountability before the most capable models are released.

Open questions

Enforceability, antitrust waivers, global verification, and incumbent advantage remain unresolved.

Continue reading

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