Just two weeks after scrapping a previous executive order on AI, President Donald Trump signed a new one — and it signals a notable shift in how the White House approaches AI governance. The order, titled “Promoting Advanced Artificial Intelligence Innovation and Security,” attempts to balance innovation with oversight, though it’s likely to draw criticism from all sides.
Here are the five key components you need to know.
1. Voluntary Review System
Tech companies will be asked — but not required — to submit frontier AI models to the government for review 30 days before public release. This is a notable step back from the previous order, which had requested 90 days of notice. The voluntary nature means compliance depends entirely on industry cooperation.
2. No Mandatory Licensing
The government will not require permits before AI tools can be deployed. This hands-off approach on licensing keeps the regulatory burden low for startups and established players alike, but critics argue it leaves dangerous capabilities unaddressed until after they’ve already reached the public.
3. AI Cybersecurity Clearinghouse
The order establishes a dedicated hub for AI security coordination, designed to run safety assessments in partnership with the private sector. This is the most concrete new infrastructure created by the order and could prove valuable regardless of one’s stance on regulation.
4. A Watered-Down Version
The new order is notably less stringent than the one Trump shelved the previous month. That earlier version mandated 90-day pre-release review and included stricter requirements. The shift reflects ongoing tension within the administration between AI safety advocates and those pushing for minimal regulatory friction.
5. A Step Toward Stronger Oversight
Despite its softer touch, the order marks a departure from the White House’s previously hands-off posture toward AI governance. The very existence of a federal review framework — even a voluntary one — represents a recognition that AI oversight is no longer optional.
Meanwhile, the defense-tech landscape is also evolving rapidly. Anduril and Meta have shared new details about their augmented-reality headset prototype for the military, which includes vision for ordering drone strikes through eye-tracking and voice commands — a glimpse into how AI is reshaping warfare as well as civilian life.
