The White House is in advanced talks with AI companies to unveil voluntary standards for frontier model releases as soon as this week, according to the Financial Times and Reuters. The announcement would implement Section 3 of President Trump's June 2 executive order on AI innovation and security, with a formal August 1 deadline looming.
What the framework will include
The voluntary standards are expected to establish three core components:
1. Classified benchmarks for designating a model as a "covered frontier model" — the threshold that triggers a voluntary 30-day pre-release government review. The benchmarks are classified; frontier labs will not know the precise criteria until they enter the framework.
2. The mechanics of the 30-day review — what materials AI companies provide to government evaluators, confidentiality protections, and who serves as government reviewers. The NSA, CISA, Treasury, and Department of War are leading the process.
3. International access rules — clarifying which foreign organizations and individuals can access covered frontier models inside and outside the United States. This directly addresses the root cause of the Fable 5 export control episode, where a jailbreak discovery led to a 19-day global model suspension.
Why now
The urgency comes from three converging pressures. First, the Fable 5 crisis — the 19-day export control suspension in June — demonstrated what happens without a clear framework: a borderline jailbreak finding triggered an emergency government response with no calibrated escalation process.
Second, Google has confirmed it is in government talks ahead of Gemini 3.5 Pro's planned July release, which sources describe as having advanced coding capabilities approaching covered frontier model thresholds.
Third, OpenAI's GPT-5.6 Sol, Terra, and Luna remain limited to approximately 20 government-vetted partners, with broader access contingent on the framework's completion. Analyst consensus places GPT-5.6's public rollout in the July 7–21 window if the standards are announced this week.
Voluntary in name, required in practice
While the framework is nominally voluntary, the Fable 5 precedent makes clear that any frontier lab without a pre-release government relationship risks emergency export controls when new capabilities emerge. Anthropic, OpenAI, Google, and Microsoft have all participated in shaping the standards.
"These rules should be codified in strong regulation and applied equally across frontier model developers," Anthropic stated after the Fable 5 restoration, signaling industry alignment with the administration's direction.
The announcement, expected as soon as July 7, will mark the most significant US AI governance framework since the Biden administration's 2023 Executive Order — and the first to directly address the tension between frontier AI capability and national security in the Trump era.
