Overview
- On Friday, June 26, reports said the U.S. government asked OpenAI to limit the initial rollout of GPT‑5.6 to a short list of vetted partners rather than a full public launch.
- The requested preview would give officials the chance to approve access customer‑by‑customer and use enterprise platforms such as Amazon Bedrock to host the model during that period.
- Earlier this month the government ordered Anthropic to cut off foreign access to its top models, Mythos 5 and Fable 5, after tests and jailbreaks showed the systems could find software and system vulnerabilities.
- A recent executive order requires agencies and companies to build a voluntary 60‑day framework that could allow government review of 'frontier' models for up to 30 days before release.
- The interventions are already shaping business plans: OpenAI is reported to be weighing a later IPO date and companies warn that tighter controls could limit defenders’ access to tools used to find and fix security flaws.