OpenAI will publicly launch GPT-5.6, its most advanced AI model, on Thursday, following a delay last month. This postponement was prompted by U.S. government requests over heightened national security concerns. These concerns centered on the potential misuse of powerful AI technologies by hostile actors.
Axios reported that the Trump administration approved a broad launch of GPT-5.6. This approval came after additional testing and a series of meetings between the company and government officials. OpenAI had limited GPT-5.6 access to a small group of vetted partners, whose details were shared with the authorities during this period of scrutiny. The White House and the U.S. Department of Commerce did not respond to a Reuters request for comment outside regular business hours.
OpenAI announced in an X post late on Tuesday that it would launch its most capable model, GPT-5.6 Sol. Lower-cost Terra and Luna models will also be released. OpenAI had previewed these models in late June, showcasing GPT-5.6 Sol's competitive performance against Anthropic's Mythos Preview on the ExploitBench cybersecurity benchmark. The company touted improved agentic capabilities across coding, biology, and cybersecurity, areas critical for both defense and potential exploitation.
Securing Western Innovation
The United States and China are locked in a critical race to develop cutting-edge AI models. Experts have warned that these advanced technologies could dramatically accelerate sophisticated cyberattacks. Such attacks could target sectors relying on complex, interconnected, and often decades-old technology systems, posing a significant threat to national infrastructure. Washington has increased its scrutiny of advanced AI model releases. This is to identify potential threats from the military or intelligence establishments in China, Russia, and other adversarial countries.
Chinese authorities have also held meetings with top tech firms. Their discussions focused on potentially restricting overseas access to China's most advanced AI models, including those yet to be released. This mirrors the global recognition of AI's dual-use nature and the imperative for national control. U.S. President Donald Trump signed an executive order establishing a voluntary framework. Under this framework, AI developers could provide covered frontier models to the U.S. government for up to 30 days before releasing them to trusted partners. This proactive measure aims to mitigate risks before widespread deployment.
The Global AI Arms Race
OpenAI competitor Anthropic abruptly disabled its most advanced AI models, Mythos 5 and Fable 5, for all users after the U.S. government's June 12 export control order. This order was issued over national security concerns. The curbs were lifted last week after Anthropic implemented certain safeguards, demonstrating the government's commitment to securing these powerful tools. While Washington has lifted export controls for Anthropic's Fable model, Mythos, specifically designed for cybersecurity professionals, remains available only to some trusted U.S. organizations. This restriction highlights the strategic importance of such capabilities.
In China, authorities are worried about the potential for Mythos to exploit software vulnerabilities. They also fear the U.S. might deploy the model against Beijing's interests. Anthropic has warned that it was probably impossible to make any AI model fully robust against jailbreaks, underscoring the inherent risks. Billionaire Elon Musk, whose SpaceXAI rivals Anthropic and OpenAI, announced on Wednesday that his company was also making its leading model Grok 4.5 available to the public. This further intensifies the global competition for AI dominance, a competition with profound implications for civilizational security.