NewsPublished 29 Jun 2026

OpenAI and Anthropic Limit New AI Models to Trump-Approved Customers During Cybersecurity Review

Source SecurityWeek

Description

ChatGPT maker OpenAI said Friday it is restricting the release of its new artificial intelligence model at the request of President Donald Trump’s administration. The post OpenAI and Anthropic Limit New AI Models to Trump-Approved Customers During Cybersecurity Review appeared first on SecurityWeek .

Impact

The restricted release limits access to advanced AI models capable of identifying software vulnerabilities, potentially reducing the risk of their misuse by malicious actors. However, the government-imposed limitations may delay broader availability and innovation. The vetting process introduces a new layer of oversight for AI products with cybersecurity implications, potentially affecting market dynamics and company operations. There are no known exploits in the wild related to these AI models. The impact is primarily on access control and regulatory oversight rather than a direct technical vulnerability or exploit.

Defensive guidance

This is a government-mandated access restriction rather than a vulnerability requiring patching. No direct remediation or patch is applicable. Organizations should monitor official communications from OpenAI, Anthropic, and relevant government agencies for updates on model availability and compliance requirements. No additional mitigation actions are necessary at this time.

Summary

OpenAI and Anthropic have limited the release of their latest AI models to customers approved by the Trump administration as part of a cybersecurity review process. This follows concerns that Anthropic's Mythos model demonstrated capabilities to find software flaws that could be exploited maliciously. The U.S. government established a framework to vet advanced AI systems for national security risks up to 30 days before public release, leading to restricted access and temporary takedowns of some Anthropic models. OpenAI's GPT-5.6 Sol model is currently available only to about 20 approved customers. Both companies emphasize that these measures are temporary and part of a phased release with stronger safeguards. The government continues to collaborate with AI labs to address risks associated with scaling AI technologies.

References

securityweek.com/openai-and-anthropic-limit-new-ai-models-to-trump-approved-customers-during-cybersecurity-review/ (opens in a new tab)

Community reviews

Loading reviews…

Description

AI analysisMachine-generated, updated 29 Jun 2026, 10:21 UTC

References

Community reviews