Story perspectives

Draft Order Splits AI Safeguards: Voluntary Vetting, Cybersecurity Push

5/21/2026

Summary

Story summary

The executive order draft splits AI safeguards into a cybersecurity push and a voluntary frontier-model vetting scheme, raising concerns about weak protection.
It would require frontier AI developers to submit models for voluntary review within 90 days.
Earlier drafts called for mandatory vetting by the Center for AI Standards and Innovation, a stricter approach.
The cybersecurity portion aims to harden federal infrastructure against AI-related threats rather than regulate model development.

1 of 1

Story summary

The executive order draft splits AI safeguards into a cybersecurity push and a voluntary frontier-model vetting scheme, raising concerns about weak protection.
It would require frontier AI developers to submit models for voluntary review within 90 days.
Earlier drafts called for mandatory vetting by the Center for AI Standards and Innovation, a stricter approach.
The cybersecurity portion aims to harden federal infrastructure against AI-related threats rather than regulate model development.