What happened
President Donald Trump cancelled the signing of a planned executive order on Thursday. The order would have granted the US government power to test frontier AI models before their public release. According to a report from Ars Technica, the event was cancelled just hours before it was scheduled after top AI firm CEOs declined to attend on 24 hours' notice. Trump told reporters he felt the order could have been an innovation 'blocker'.
How the room's reading it
The cancellation has split opinion among AI labs. Reporting suggests OpenAI supported the order, while Meta's Mark Zuckerberg and xAI's Elon Musk — who denied his involvement — reportedly lobbied against it. The core tension seems to be about pre-release testing timelines. The government wanted up to 90 days to vet models for security risks, a timeline that infra teams and labs feared would significantly delay launches and set back development. The consensus among practitioners is that the industry pushback was successful, framing the order as a potential brake on innovation in the race against China.
Sailfish's take
We see this as a temporary reprieve, not a permanent win for a 'light-touch' approach. The underlying security concerns that prompted the order haven't disappeared — Anthropic's own flagging of risks with its Mythos model proves that. We're advising our portfolio companies to continue building robust internal safety and testing programmes. Relying on last-minute lobbying to avoid regulation is a fragile strategy. The useful question isn't whether regulation is coming, but what sensible, builder-friendly regulation looks like. This episode just kicked the can down the road.