OpenAI is expected to launch its new series of GPT-5.6 artificial intelligence models on Thursday, after the release was delayed for several weeks due to US authorities' concerns over national security.

According to Axios, the White House approved the release only after technical testing and consultations with company executives. The delay was due to fears that advanced artificial intelligence models could be used to identify vulnerabilities in software systems that could be exploited by rival states such as Russia and China.

OpenAI calls GPT-5.6 the most advanced model developed to date, with improved capabilities in cybersecurity, biology, and autonomous task performance.

The GPT-5.6 series includes three different versions: Sol, the company's most powerful model; Terra, designed for everyday use; and Luna, the fastest and most economical version.

The company also announced that GPT-5.6 has been equipped with a new multi-layered security system, designed to prevent the misuse of artificial intelligence in cyberattacks and other dangerous activities.

Although it accepted the delay requested by the White House, OpenAI stressed that it does not want government prior scrutiny to become a permanent practice, arguing that this could delay users and businesses' access to cutting-edge technologies.