In a move that has left the tech community reeling, OpenAI and Anthropic have both been restricted by the Trump administration from releasing their latest AI models to the public due to concerns over national security risks. The unprecedented government vetting of these powerful technologies has sparked debate about the balance between innovation and regulation.
The saga began earlier this month when Anthropic’s new AI model, Mythos 5, was temporarily blocked by the government after it was deemed a potential cyber threat. The company complied with the directive, taking the model offline, but not before warning that its capabilities could be exploited by malicious hackers to compromise critical computer networks worldwide. Now, OpenAI has announced that its own new AI product, GPT-5.6 Sol, will only be available to customers approved by the Trump administration.
The restrictions have been met with criticism from both within and outside of the tech industry. Some argue that unpredictable government intervention can stifle innovation and hold back US companies in the global market. “No law. No process. No oversight. Just appointees in Washington deciding who’s in and who’s out,” said U.S. Rep. Lori Trahan, a Massachusetts Democrat and co-author of a bipartisan bill to regulate AI.
But what exactly is driving these concerns? New, powerful AI models like Mythos 5 have the potential to be used for both good and ill. On one hand, they can help identify software flaws and improve cybersecurity defenses. However, in the wrong hands, their capabilities could be exploited to launch devastating cyberattacks. OpenAI claims that its new Sol model is designed to prioritize finding vulnerabilities over carrying out attacks, but even it acknowledges that there may be unforeseen risks if combined with other tools.
The government’s actions have also raised questions about the effectiveness of its regulatory framework for AI oversight. The Trump administration signed an executive order in June establishing a 30-day review period for the most advanced AI systems before their public release. However, this framework is still in development and has yet to be fully implemented.
Meanwhile, Anthropic’s CEO Dario Amodei has been criticized by some of Trump’s allies for allegedly creating a “cyber weapon” with Mythos 5. While Amodei has maintained that the model was designed to improve cybersecurity defenses, its capabilities have sparked concerns about potential misuse. As the debate rages on, one thing is clear: the development and deployment of powerful AI technologies will require careful consideration of both their benefits and risks.
For now, OpenAI’s Sol model remains restricted to a small group of trusted partners, while Anthropic’s Mythos 5 has been redeployed to a select few cyber defenders and infrastructure providers. As the tech industry grapples with these complex issues, one thing is certain: responsible AI development will require collaboration between government, industry leaders, and experts in the field.
Source: SecurityWeek — 2026-06-29