OpenAI To Launch GPT-5.6 After US Government Security Review

OpenAI is set to release its GPT-5.6 AI model after the US government completed a security review that delayed its launch. The company says the new model offers stronger capabilities, improved safety features and will be available in three versions, while continuing to work with US authorities on future AI oversight.

OpenAI is preparing to release its new GPT-5.6 artificial intelligence model after receiving approval from the United States government following weeks of security reviews.
The company had originally planned to launch the model in June but delayed the release after US officials requested more time to assess whether the advanced technology could pose national security risks.
According to reports, the White House has now completed its review and cleared the release after holding discussions with OpenAI executives and carrying out technical assessments.
OpenAI says GPT-5.6 is its most advanced AI model so far, offering improved performance in areas such as cybersecurity, scientific research and autonomous AI systems that can complete complex tasks with limited human supervision.
The launch comes as governments around the world continue debating how to balance rapid AI development with national security and public safety.

US officials were concerned that increasingly powerful AI systems could potentially be misused if they fell into the wrong hands.
The review focused mainly on the model's ability to analyse software, identify security weaknesses and perform advanced technical tasks.
Following weeks of testing and discussions, the US government approved the public release.

Officials worry that countries such as Russia and China could potentially use powerful AI tools to improve cyberattacks or gather intelligence.
Because of these risks, the US government has increased its oversight of the most advanced AI models before they become publicly available.
Experts say balancing innovation with national security has become one of the biggest challenges facing governments.

However, the company also said it does not believe government access before every AI release should become the normal approach.
According to OpenAI, delaying powerful AI tools for long periods could prevent businesses, researchers and the public from benefiting from new technology.
The company said it wants future reviews to follow a clear and consistent process.

The most powerful version, called Sol, is designed for highly complex tasks requiring advanced reasoning and problem-solving.
A second version, Terra, is intended for everyday professional and business use.
The third version, Luna, is designed to provide faster responses while remaining the most affordable option.
The company says each version is designed to meet different user needs.

The company described these protections as a "layered safeguard system" designed to reduce the risk of cybercrime, harmful content and other dangerous activities.
Developers say the model has been trained to better recognise unsafe requests while maintaining strong performance for legitimate users.
The company added that safety testing will continue even after the public release.

Although President Donald Trump has generally supported fewer regulations to help the United States compete with China in artificial intelligence, his administration has also introduced measures to address cybersecurity risks linked to advanced AI systems.
OpenAI said it will continue working with US authorities to develop a long-term process for reviewing future AI models before they are released.
Industry experts believe similar cooperation between governments and AI companies is likely to become increasingly common as artificial intelligence continues to advance.