Trump Signs Diluted AI Oversight Order Following Silicon Valley Pushback
Table of Contents
A Strategic Retreat on AI Regulation
President Donald Trump signed an executive order on Tuesday that establishes a framework for government review of high-capacity AI models, but the final text reveals a significant concession to the tech industry. The order asks developers of powerful AI systems to voluntarily submit new models for government testing or evaluation 30 days before they hit the public market—a compromise that reflects a tense tug-of-war between national security hawks and Silicon Valley’s fastest-moving firms.
The final 30-day window is a notable retreat from earlier drafts. Internal versions of the order had proposed a more stringent 90-day review period, a timeline that industry insiders argued would stifle innovation and hand a competitive advantage to adversaries. While some lobbyists pushed for a window as short as two weeks, the administration landed on a month, attempting to balance the need for safety benchmarks with the urgency of the current AI arms race.
The shift in timing suggests that industry influence played a decisive role in the drafting process. David Sacks, a prominent venture capitalist and former White House AI advisor, was among the voices pushing back against the more demanding version of the order. Trump, who had delayed the signing of the original May draft, explicitly stated his reluctance to hinder American AI firms in their struggle to maintain a lead over China.
The ‘Non-Mandatory’ Clause
Perhaps more critical than the timeline is the specific language used to prevent the order from becoming a regulatory bottleneck. The document explicitly states: “Nothing in this section shall be construed to authorize the creation of a mandatory governmental licensing, preclearance, or permitting requirement for the development, publication, release, or distribution of new AI models, including frontier models.”
By framing the review as voluntary and explicitly denying the government’s power to mandate licensing, the administration has avoided creating a formal “regulatory gate.” For companies like OpenAI, Anthropic, and Google, this ensures that while the government may be kept in the loop, the White House cannot legally block a product launch based on these specific reviews.
The optics of the signing also shifted. Originally, the event was envisioned as a high-profile gathering featuring a roster of Silicon Valley’s top CEOs. Instead, Trump signed the current version privately, a move that suggests a desire to avoid the public spectacle of a scaled-back policy or perhaps a reflection of the fractured relationship between the administration and certain tech leadership.
Beyond Model Reviews: Cybercrime and Federal Preemption
The executive order extends beyond model evaluation, signaling a more aggressive stance on the misuse of artificial intelligence. It specifically directs the Department of Justice (DOJ) to prioritize the enforcement of crimes involving AI-assisted hacking and unauthorized network access. This pivot acknowledges a growing reality: as LLMs become more capable of generating polymorphic code and automating phishing campaigns, the DOJ needs a specialized mandate to target AI-driven cyber warfare.
This order is the latest piece of a broader puzzle. In December, Trump signed a separate order aimed at creating a “one rulebook”—a national AI policy framework. That initiative is designed to preempt a patchwork of contradictory state-level AI laws, effectively attempting to federalize AI regulation to ensure a uniform standard for businesses operating across state lines.
By combining this voluntary review process with a push for federal preemption, the administration is attempting a delicate balancing act: providing enough oversight to appease safety advocates and national security officials, while ensuring that the bureaucratic machinery does not slow the deployment of American technology.
