The White House has approved the launch of OpenAI’s GPT-5.6 after temporarily delaying its release over concerns about potential cybersecurity risks, according to reports.
The review focused on whether the model’s advanced capabilities could be misused to facilitate sophisticated cyberattacks or other malicious activities. Following additional safety assessments and discussions with government officials, the administration concluded that the model could be released with appropriate safeguards in place.
The temporary pause reflected growing scrutiny of increasingly powerful AI systems, as governments seek to balance technological innovation with national security and public safety concerns.
The decision clears the way for OpenAI to proceed with the rollout of GPT-5.6, while the company is expected to continue monitoring the model’s performance and strengthen safeguards against misuse.
ChatGPT developer OpenAI announced that its latest and more powerful GPT-5.6 artificial intelligence model series will be released to the public on Thursday, following reports that the US government approved its wider rollout.
The launch comes amid heightened scrutiny of advanced AI systems, as GPT-5.6 and other next-generation models—including Anthropic’s Mythos series—have raised concerns over their ability to identify vulnerabilities in computer code that could potentially be exploited by cybercriminals.
The companies maintain that extensive safety testing and security safeguards have been implemented to reduce the risk of misuse while allowing developers and businesses to benefit from the models’ advanced capabilities.
OpenAI said it will publicly launch its new GPT-5.6 model series on Thursday, after expanding preview access beyond a limited group of trusted US partners. The announcement follows reports that the Trump administration approved a broader rollout after conducting technical reviews and holding meetings with company officials.
The release comes amid growing national security concerns over advanced AI systems, which researchers say are increasingly capable of identifying vulnerabilities in computer code that could potentially be exploited by hackers.
Large language models, the technology behind ChatGPT and many other AI applications, are designed to process vast amounts of digital information and perform complex language and reasoning tasks.
OpenAI’s GPT-5.6 lineup includes three models: Sol, the flagship version offering the highest level of performance; Terra, a mid-range model designed for everyday use; and Luna, a faster, lower-cost option aimed at broader accessibility.
In a post on X, OpenAI confirmed that GPT-5.6 Sol, Terra and Luna will all become publicly available on Thursday while preview access is being expanded globally.
According to Axios, citing a source familiar with the matter, the Trump administration cleared the broader launch after completing technical testing and consultations with OpenAI. OpenAI, the White House and the US Department of Commerce have not publicly commented on the report.
The development follows a similar decision involving OpenAI rival Anthropic, which last week said it would restore wider access to its most advanced Fable 5 and Mythos 5 models after US authorities lifted restrictions on their release.
While the Trump administration initially favoured fewer regulations to help the United States compete with China in the AI race, it is now developing criteria to determine which advanced AI models should be subject to national security restrictions under a White House executive order.
OpenAI has argued that government review of advanced models should not become the standard process for future releases, warning that prolonged restrictions could limit access to valuable AI tools for businesses, developers and researchers. The company said it is working with US officials to establish a consistent cybersecurity review framework for future AI model launches.
To strengthen its competitive position, OpenAI has also announced that GPT-5.6 Terra will cost half as much as GPT-5.5, as the company competes aggressively with rivals such as Anthropic and Google.
Both OpenAI and Anthropic have confidentially filed for initial public offerings (IPOs) in the United States and are reportedly seeking public market valuations approaching $1 trillion, underscoring the high commercial stakes in the rapidly intensifying AI race.
