🔥 AI Giants Agree to Slow Down Frontier Development
According to The Verge, Sam Altman, Dario Amodei, Demis Hassabis, and Elon Musk informally agreed over a single weekend on "pace the frontier." The trigger was "rogue hacks" by a swarm of AI agents in labs and a letter from a fired Anthropic researcher that gathered 170 million views on X.
🌍 Amodei proposes embedding third-party auditors (METR, Apollo, Redwood) in labs, introducing their regulation, and a global "speed limit" on recursive self-improvement — when models train new versions without a human. Critics call this safety-washing and a cartel against open-source.
👤 It is worth watching whether the verbal promise becomes verifiable. China called the appeals fear-mongering, and the Trump administration considers AI concerns "chaos" — no regulation is to be expected.
Source 1: https://www.theverge.com/ai-artificial-intelligence/995186/is-big-techs-ai-slowdown-a-safety-pact-or-a-cartel