What happened
President Donald Trump has delayed signing an executive order on AI security. According to TechCrunch, the order would have required government evaluation of new AI models before their public release.
The draft text reportedly included a provision for companies to share advanced models with the government between 14 and 90 days ahead of launch. Trump cited concerns that the language "could have been a blocker," stating he didn't want to hinder the US lead in AI.
How the room's reading it
The official line frames the delay as pro-innovation — a move to avoid stifling the US lead in AI. Policy watchers, however, are pointing to reports that the signing was postponed simply because not enough tech CEOs could attend a photo op.
For security practitioners, the delay is a mixed bag. The proposed order was a direct response to concerns over models capable of finding and exploiting vulnerabilities, so its postponement leaves a perceived gap. Meanwhile, many developers see the proposed pre-release government reviews as a significant bottleneck, viewing any delay as a welcome reprieve from burdensome compliance.
Sailfish's take
We're reading this as a temporary stay, not a change in direction. The political appetite for regulating powerful AI models isn't going away — it's just been postponed for a photo op. This isn't a win for deregulation; it's a brief window for builders to ship without immediate federal oversight.
Our advice is to use this time wisely. Don't mistake the delay for a green light to be reckless. Instead, this is the moment to build and ship products with robust internal safety and security practices. Proving the industry can self-regulate effectively is the only real way to shape the inevitable rules to come.