Washington: New artificial intelligence tools, particularly advanced AI models, developed by major tech firms including Google, Microsoft, and xAI, will now be tested by the US Department of Commerce before being released to the public.
The companies have voluntarily agreed to submit each AI model for evaluation through the Department’s Center for AI Standards and Innovation (CAISI). This initiative expands on earlier agreements reached during the Joe Biden administration with firms such as OpenAI and Anthropic.
Under the new framework, every AI model will be assessed for performance, safety, and potential risks. Chris Fall, Director of CAISI, said that the expanded partnerships will help scale efforts to evaluate each AI model in the public interest at a critical stage of technological development.
AI models under review
The programme will focus on testing, collaborative research, and the development of best practices for deploying commercial AI models. Among the systems under review are;
- Google’s Gemini, an advanced AI model developed by its DeepMind division, which is widely integrated across products and increasingly used in US defence operations.
- Microsoft’s Copilot is another widely used AI model.
- xAI’s Grok, also an AI mode, has faced public scrutiny over issues such as generating inappropriate altered images.
CAISI confirmed that it has already conducted around 40 evaluations of various AI models, including several state-of-the-art AI models that have not yet been released to the public. However, the agency did not specify which AI models were withheld.
In a corporate statement, Microsoft noted that it already conducts internal testing of each AI model, but emphasised that evaluating national security and large-scale safety risks for any AI model requires close collaboration with government authorities.
The move to expand testing across companies signals a shift from the previously hands-off regulatory stance under Donald Trump. His administration had introduced an ‘AI Action Plan’ aimed at reducing regulatory barriers and accelerating the development of each AI model to maintain US leadership in the sector.
However, with the US military increasing its reliance on advanced AI models, and recent claims by Anthropic that it developed an AI model called Mythos considered too powerful for public release, the administration’s position appears to be evolving.
Senior officials from the Trump administration recently met with Anthropic CEO Dario Amodei to discuss concerns surrounding the deployment of highly advanced AI models. This comes as Anthropic faces a legal dispute with the US Department of Defense over its refusal to remove safety guardrails from its systems for government use.
ALSO READ | Tesla pushes for EU-wide approval of FSD in Brussels meeting
