OpenAI is set to introduce its latest and most sophisticated AI model, named GPT-5.6, on Thursday. The introduction of the AI model has been postponed recently due to national security fears raised by the U.S. government about potential misuse of powerful AI technology.
There is a race between the United States and China to develop the most sophisticated AI system. Specialists claim that such AI models might become tools for performing complicated cyberattacks against sectors dependent on outdated and interrelated technology.
Thus, the U.S. government carefully examines advanced AI models before their introduction due to the risk that other countries, such as China and Russia, can make use of the technology for military and intelligence purposes.OpenAI has released its new AI models after getting the approval from the US government due to some additional testing and discussions regarding national security issues.
Prior to the release, GPT-5.6 became accessible to some trusted partners whose information has been reported to the US government.
Three new models have been introduced by OpenAI:
GPT-5.6 Sol is the most developed one.
Terra is a relatively inexpensive model.
Luna is also an inexpensive model.
OpenAI states that these models demonstrate superior performance in coding, biology, cybersecurity, and AI agents’ jobs. During tests, the performance of GPT-5.6 Sol was similar to Anthropic’s Mythos Preview in the field of cybersecurity.On the other hand, billionaire Elon Musk has revealed that his company has also launched the latest version of their AI model called Grok 4.5 in the public domain.
The United States government has come up with a voluntary scheme where AI firms can provide the powerful AI model to the government for 30 days before launching it publicly.
Previously, Anthropic restricted access to its AI models Mythos 5 and Fable 5 due to new regulations by the United States regarding exports control. However, access to Fable was provided again after Anthropic made safety features in it, while access to Mythos is limited only to some select US organizations because of their cybersecurity.
There are also plans by the Chinese government to restrict access to AI models in China due to concerns about the use of these models for finding weaknesses in software or posing threats to national security. Anthropic has also stated that it is extremely hard to make any AI model safe from being jailbroken.
