Critics Dismiss AI Safety Slowdown as ‘Too Little, Too Late’

Plans proposed by Anthropic chief to enhance artificial intelligence safety have triggered widespread criticism from multiple quarters, including the Trump administration and various AI specialists. The initiative, which calls for a slowdown in development due to concerns over reckless progress, has been met with suspicion and confusion rather than support.

The rapid reversal in public sentiment highlights the volatility of the current safety debate. Just one week prior, enthusiasts were enthusiastically engaging with OpenAI’s latest model, GPT-6 Astra, which was promoted as a versatile lifestyle tool for tasks ranging from booking tennis courts to managing fashion businesses and ordering food.

This enthusiasm shifted sharply by Wednesday following warnings from Jacob Coxon, a resigning Anthropic researcher. Coxon stated that those building AI earnestly believe the technology could pose an existential threat to humanity by the end of the decade. Although his comments referred to artificial super-intelligence—a hypothetical iteration of the technology that surpasses human capability across most domains—they resonated with a growing chorus of experts, causing significant unease.