A look at a copy of the AI EO that President Trump was expected to sign on May 21; the unsigned EO emphasized that government AI reviews would be voluntary
The draft also includes language aimed at bad actors. It directs the attorney general to enforce the Computer Fraud and Abuse Act and …
Context & Ripple Effects
The draft followed earlier moves to remove AI-related policies characterized as barriers to innovation, while the White House was also considering measures focused on advanced-AI security risks and government access to models.
Related coverage shows the eventual order was scaled back and focused on cybersecurity, with less AI scrutiny than the abandoned draft. That makes the voluntary-review language a useful marker of the administration’s preferred balance between security intervention and lighter oversight.
First-order effects
- AI companies would face no mandatory government review process under the unsigned draft’s approach, while remaining exposed to enforcement of existing computer-fraud law against misuse.
- The immediate federal emphasis shifts toward targeting identified bad actors and cybersecurity threats rather than creating a broad pre-deployment AI review regime.
Second-order effects
- Voluntary reviews give leading developers discretion over whether and how to engage federal evaluators, reducing a near-term compliance burden but also limiting a common government baseline for assurance practices.
- Competitors and enterprise customers may increasingly differentiate AI systems through self-directed security commitments, because the draft does not establish a compulsory review standard.
Third-order effects
- If the signed, scaled-back approach persists, US AI governance is likely to rely more on existing enforcement authorities and targeted cybersecurity actions than on a centralized, mandatory model-oversight framework.
- That leaves AI assurance practices more dependent on company choices and sector-specific requirements, potentially producing uneven safeguards unless later federal action sets firmer common standards.
The trend: This is one data point in a shift toward AI policy that prioritizes innovation and targeted security enforcement over broad, mandatory federal model review.