OpenAI will delay the launch of its next big model to ChatGPT, GPT-5.6after the White House have requested limit AI distribution to a small group of government-approved partnersinstead of opening it to the general public as she has done with previous releases, concerned about possible security issues.
According to The Information, the CEO of OpenAI, Sam Altmansent an internal note to the staff on Thursday in which he explained that Federal Government to approve customer-by-customer access during initial preview phase. Altman noted in that communication that this restrictive deployment ‘is not (its) long-term model’ for distributing software and that the company will later work on a ‘more sustainable’ method.
Other information from CNN states that Both OpenAI and the Trump administration place GPT-5.6’s capabilities on the same level as Anthropic’s Mythosand that government officials intend to ‘collaborate with the most advanced AI laboratories to develop common approaches to meet the challenges of scaling this technology.’
Anthropic’s precedent with Mythos and Fable
The latest restriction comes just weeks after the Department of Commerce of the United States decided restrict both the use of Mythos and Fablethe latest Anthropic models.
The first was launched in April only for a group of partners selected by its capabilities in cybersecurity and to detect vulnerabilities, which gives it a high degree of danger in the wrong hands. Fable arrived in June as a version of Mythos with additional security ‘safeguards’ to prevent users from exploiting software vulnerabilities.
However, shortly after launch, Amazon researchers found a way to bypass those restrictionswhich provoked a strong response from federal authorities. The Government ordered Anthropic cut off access to both models to non-US citizens both outside and inside the United States, including those who work at Anthropic.
Anthropic is reportedly developing a solution to resolve this blockage through an update to its privacy policy that introduces a category called ‘Verification data’ to manage digital IDs. This system could impose identity checks to filter users by nationality, requiring an official document and facial biometric data.