OpenAI to Launch GPT-5.6 for Everyone After US Delayed Release Over National Security Concerns

OpenAI is set to publicly launch GPT-5.6 on Thursday, ending weeks of uncertainty after the company delayed its rollout at the request of the US government over concerns that the model's advanced capabilities could pose national security risks.

The launch marks one of the first major AI model releases to be delayed because of government scrutiny, highlighting how frontier AI is increasingly being viewed through the lens of cybersecurity and geopolitics rather than just technological innovation.

Why GPT-5.6 Was Delayed

Last month, OpenAI postponed the public rollout of GPT-5.6 after the US government requested additional testing and consultations over fears that highly capable AI models could be misused for cyberattacks or by foreign adversaries.

According to Reuters, the Trump administration has stepped up oversight of advanced AI systems amid concerns that countries such as China and Russia could exploit frontier AI models for military or intelligence purposes.

Axios, which first reported the development, said the White House approved a wider launch only after additional meetings and safety evaluations involving OpenAI and government officials.

OpenAI Is Launching Three New Models

Alongside its flagship GPT-5.6 Sol, OpenAI will also release two lower-cost models called Terra and Luna. When previewing the models in June, the company said GPT-5.6 offers stronger agentic capabilities across coding, biology and cybersecurity. OpenAI also claimed the model performed competitively against Anthropic's Mythos Preview on the ExploitBench cybersecurity benchmark.

Before the public rollout, access to GPT-5.6 was restricted to a small group of vetted partners whose identities were shared with US authorities.

AI Releases Are Becoming a National Security Issue

The GPT-5.6 delay reflects a broader shift in how governments are approaching advanced AI.

Earlier this year, rival Anthropic temporarily disabled public access to its Fable 5 and Mythos 5 models after the US imposed export controls over national security concerns. While restrictions on Fable have since been lifted, Mythos remains available only to selected trusted organisations because of its cybersecurity capabilities.

The US has also introduced a voluntary framework under which AI developers can share frontier models with government agencies for up to 30 days before broader releases.

The AI Race Is No Longer Just About Chatbots

The launch comes as the AI rivalry between the US and China continues to intensify. Chinese authorities have reportedly held discussions with leading technology companies about restricting overseas access to their most advanced AI models, while American regulators are increasingly evaluating whether cutting-edge AI systems could accelerate sophisticated cyberattacks or expose critical infrastructure to new risks.

Competition is also heating up among AI companies. On the same day, Elon Musk announced that Grok 4.5, developed by his AI venture, would also become publicly available.

The rollout of GPT-5.6 therefore represents more than another ChatGPT upgrade. It signals that every major frontier AI model may now face government scrutiny before reaching users, making regulation and national security an increasingly important part of the global AI race.