Leaders at Anthropic, OpenAI, Microsoft and SpaceXAI called for slower development of frontier AI models. They cited a gap between what the technology can do and the safety rules around it. Anthropic chief executive Dario Amodei made the case in a weekend blog post. He proposed a three-part safety framework, including outside evaluators embedded inside the labs. Sam Altman, Satya Nadella and Elon Musk echoed the concern. Gartner analyst Arun Chandrasekaran said the shift adds uncertainty for technology leaders who had assumed a fixed release pace. Info-Tech Research Group analyst Mark Tauschek said voluntary statements are not enough. He advised adoption gates reviewed by security, privacy, legal and risk owners, plus contract rights to pause, be notified or exit. A former Anthropic researcher quit last week warning of extinction risk. President Donald Trump criticised the slowdown push as a threat to US competitiveness.

• three-part safety framework proposed