Archive — US — techcrunch
The White House is asking OpenAI to slow roll the release of its new model oversafety concerns
The White House is urging OpenAI to limit the public release of its new model, GPT 5.6, due to safety concerns, only allowing access to selected partners. This shift is prompted by the Trump administration's call for strict federal oversight of AI technologies, reflecting a trend towards cautious deployment as the risks of generative AI are becoming increasingly apparent.
Key points
- OpenAI's GPT 5.6 will have a limited release due to safety concerns raised by the Trump administration.
- CEO Sam Altman confirmed that access to the model will be approved on a case-by-case basis.
- The limited release strategy mirrors a previous approach taken by Anthropic with its model Claude Mythos.
- Federal agencies like the Office of the National Cyber Director have requested this cautious deployment.
- Trump's executive order necessitates AI companies to submit models for government evaluation before public release.
- Concerns arise about the potential misuse of powerful AI models in cybercrime, including automated malware and ransomware attacks.
Why it matters
- The trend towards increased regulation in AI mirrors past instances where emerging technologies faced government scrutiny for public safety, such as in the early internet age.
- The rise in generative AI's capabilities has highlighted a need for responsible development, reminiscent of the debates on social media's impact on society.
- The cautious approach by OpenAI and Anthropic reflects a growing industry recognition of the risks posed by advanced AI, indicating a shift in how companies manage AI's power.
The limited release of GPT 5.6 signifies a pivotal moment for AI governance, as companies prioritize safety and oversight amid rising concerns about the implications of advanced AI technologies, shaping the future direction of the field.