AI 'Pacing' Proposal Falls Short of Addressing Extinction Risk
Experts warn Anthropic's frontier-pacing plan offers a one-to-two-year slowdown but no credible path to controlling advanced AI systems.
A growing chorus of AI safety researchers argues that a proposed industry slowdown championed by Anthropic CEO Dario Amodei — and endorsed by Sam Altman and Elon Musk — does not adequately address the existential risks posed by rapidly advancing artificial intelligence systems.
The debate intensified following the resignation of Jacob Coxon from Anthropic, an event that AI safety advocates say has pushed concern over AI risk into mainstream public consciousness. For years, experts have warned that AI companies are advancing capabilities without a reliable method of maintaining control over the systems they build. Coxon's departure, observers say, signals that those internal concerns are no longer containable.
Read more UK Unions Praise Labour Team but Brace for Rights Battles →
Amodei's "pacing the frontier" framework envisions a slowdown of roughly one to two years, during which the industry would pursue what he describes as "profound progress" on technical safety measures. Critics, however, say that timeline is insufficient given the scale of the challenge. The stark underlying problem, safety researchers contend, is that no one currently knows how to guarantee control over advanced AI systems — and a brief pause does not resolve that fundamental gap.
What is needed, according to those critics, is not a temporary deceleration but a concrete, credible plan demonstrating that advanced AI cannot cause catastrophic harm. Without such a framework, any resumption of frontier development after a short pause would leave core dangers unaddressed. The proposal, as currently framed, offers reassurance without the technical foundation to justify it.
Continue reading at Business | The Guardian