Tech Giants Share AI Models with Government for Cybersecurity Review

New York. To prevent cybersecurity challenges that can arise from Artificial Intelligence (AI), world-renowned technology companies Google, Microsoft, and XAI have agreed to hand over their unreleased new AI models to the government. The National Institute of Standards and Technology (NIST) of America officially announced this partnership on Tuesday.

This major decision is a result of the debate that started last month after the powerful new AI model 'Mythos' developed by Anthropic company came into the market. Concerns about the serious impact Mythos could have on cybersecurity reached their peak, forcing the White House to consider starting a formal review process for AI models.

Under the new agreement, the Center for AI Standards and Innovation (KC) under the US Department of Commerce will evaluate new AI models before they are released to the market. This institution has been given the authority to conduct a detailed study of the potential impact of these models on national security and public safety. In addition, this center will continue research and testing even after the AI models are released to the market. This center has successfully evaluated more than 40 AI models so far.

KC Director Chris Fall said in a statement, 'Independent and rigorous scientific measurement is necessary to understand frontier AI and its impact on national security. This expanded collaboration with industries will help us increase the scope of working for public interest at this important juncture.'

According to Anthropic company, 'Mythos' is far ahead of other existing models in terms of cybersecurity. This is the reason why there was a great security concern among governments, banks, and service provider companies of various countries regarding this technology last month.

The company has also stated that it would not be safe to make this model publicly available immediately. Currently, its access is given only to a limited group of approved institutions, and the company has already informed senior US government officials about its capabilities.

Similarly, another major technology company OpenAI clarified last week that it is providing its most advanced AI models to certified government bodies with the aim of preventing potential security threats through AI.

According to Jessica Gee, a senior research analyst at Georgetown University's Center for Security and Emerging Technology, such collaboration will provide more resources for AI testing to the government agency KC. She said, 'Government agencies do not have as much manpower, technical staff, and computing power as these big technology companies, which makes it difficult for them to rigorously test these models.'

The White House is currently consulting with a group of experts on how to structure the government review process for new AI models. This step is analyzed to be very different and stricter than the flexible policy that the Trump administration has been adopting in AI regulation so far.

However, a White House spokesperson commented on this matter, stating that discussions about potential executive orders are merely speculation and official policy announcements will come directly from the President.

Microsoft's Chief Responsible AI Officer Natasha Crampton expressed confidence that although the company regularly tests its models, collaboration with KC will provide additional technical, scientific, and national security expertise. Google has refused to comment further on this agreement, while XAI has also not responded so far.