US Department of Commerce to Test AI Models from Leading Tech Firms
The US Department of Commerce will now conduct safety evaluations on new artificial intelligence (AI) tools and capabilities developed by Google, Microsoft, and xAI before these technologies are released to the public.
These technology companies have voluntarily agreed to submit their AI models for testing through the Commerce Department's Center for AI Standards and Innovation (CAISI).
This initiative expands upon previous agreements made with AI companies such as OpenAI and Anthropic during the Biden Administration. Under this expanded framework, AI models from all participating companies will be assessed for their capabilities and security measures.
"These expanded industry collaborations help us scale our work in the public interest at a critical moment," said Chris Fall, director of CAISI.
The evaluations will encompass testing, collaborative research, and the development of best practices related to commercial AI systems.
Overview of AI Tools from Participating Companies
Google's prominent AI tool is Gemini, developed by its DeepMind subsidiary. Gemini is a chatbot widely integrated into Google products and is currently being utilized by US defense and military agencies.
Microsoft's well-known AI product is CoPilot, which assists users across various applications.
xAI, controlled by SpaceX and Elon Musk, offers Grok, a chatbot that has faced significant public scrutiny due to controversies involving inappropriate image processing.
Previous Evaluations and Transparency
On Tuesday, CAISI announced it has conducted 40 evaluations of AI tools to date. These assessments include testing certain "state-of-the-art models that remain unreleased." However, the center did not disclose which specific models have been withheld from public release.
Requests for comment from representatives of Google, Microsoft, and SpaceX, the parent company of xAI, were not answered.
Shift in US Government Approach to AI Oversight
The inclusion of additional companies for research and safety testing of commercial AI tools marks a notable shift from the previous administration's stance. The Trump White House largely adopted a hands-off approach regarding oversight and regulation of AI and technology firms.
In the prior year, then-President Donald Trump signed several executive orders forming the foundation of his administration's "AI Action Plan." This plan aimed to "remove red tape and onerous regulation" surrounding AI development and to ensure the US would "win" through technological advancements and control.
However, with the US military expanding its use of AI and recent announcements by Anthropic about a model named Mythos deemed too powerful for public release, the current White House appears to be adjusting its perspective.
Senior members of the Trump administration met last month with Anthropic CEO Dario Amodei, as previously reported by the BBC. This engagement occurred despite Anthropic being involved in a lawsuit with the US Department of Defense over the company's refusal to remove safety guardrails for government use of its AI models.
