About a week ago, the Commerce Department's Center for AI Standards and Innovation (CAISI) announced a significant agreement with major AI companies including Google DeepMind, Microsoft, and xAI. The deal allowed government inspectors to evaluate unreleased AI models before public deployment. However, the official announcement page has mysteriously vanished from the CAISI website, raising questions about transparency and accountability in the rapidly evolving field of artificial intelligence.
The agreements, dated May 5, 2026, were described as expanded industry collaborations that would enable CAISI to conduct pre-deployment evaluations and targeted research. The goal was to assess frontier AI capabilities and advance the state of AI security. Notably, Anthropic and OpenAI had signed similar agreements back in 2024, setting a precedent for this type of government oversight.
But when reporters and the public tried to access the official page using its original URL, they encountered an error message saying, "Sorry, we cannot find that page." Later, the URL redirected to the main CAISI page on the Commerce Department website. As of this writing, the redirect remains in place. The only trace of the announcement now exists in archived versions, such as those captured by the Wayback Machine.
The archived announcement stated: "These agreements support information-sharing" and emphasized "ensuring a clear understanding in government of AI capabilities and the state of international AI competition." The removal of this page without explanation has sparked speculation. Some suggest it could be a technical error, but others believe it may indicate a policy shift or a desire to recalibrate public messaging around AI regulation.
This incident occurs against a backdrop of heightened government interest in AI safety and standards. The Biden administration had previously issued an executive order on AI, and the Commerce Department has been central to developing guidelines for responsible AI development. The CAISI center itself was established to lead these efforts, focusing on standards and innovation in AI.
The deals with Google DeepMind, Microsoft, and xAI are particularly significant because they involve some of the most advanced AI models in development. Google DeepMind is a pioneer in AI research, known for breakthroughs like AlphaGo and AlphaFold. Microsoft has invested heavily in OpenAI and is integrating AI into its products. xAI, founded by Elon Musk, aims to understand the true nature of the universe through AI.
Allowing government inspection of unreleased models is a major step toward ensuring that powerful AI systems are safe before they reach the public. However, the disappearance of the announcement page undermines the transparency that such agreements are supposed to foster. Critics argue that without a permanent record, it becomes difficult for the public and researchers to hold companies and the government accountable.
The missing page also raises practical questions. What information was originally contained on the page that might now be considered sensitive? Was it removed at the request of the companies involved, or was it an internal decision? The Commerce Department and the White House have not yet responded to requests for comment. Gizmodo contacted both agencies but did not receive an immediate reply.
This is not the first time a government page related to AI policy has gone missing. Previous administrations have faced criticism for removing or altering online content without notice. The practice can erode trust and make it harder for historical analysis of policy decisions. In the fast-moving world of AI, where new developments occur daily, transparency is crucial for public confidence.
The broader context includes ongoing debates about AI regulation. Some industry leaders have called for more government oversight to prevent potential harms, while others worry that excessive regulation could stifle innovation. The United States has adopted a sectoral approach, relying on agencies like NIST and the Commerce Department to set voluntary standards. The CAISI agreements were part of this voluntary framework.
The disappearance of the page might also reflect internal tensions within the government about how to communicate AI initiatives. The Trump administration, which took office in 2025, has taken a different stance on technology regulation compared to its predecessor. The America's AI Action Plan mentioned in the announcement may have influenced the renegotiation of earlier agreements. Some observers speculate that the new administration wants to emphasize a more free-market approach, potentially downplaying the role of government oversight.
xAI's involvement adds another layer of interest. Elon Musk has been vocal about AI risks and has called for regulation. Yet his company signing a pre-deployment inspection deal suggests a willingness to cooperate with government authorities. However, if the page about the deal disappears, it could indicate that the terms of the agreement are not fully aligned with the current administration's messaging.
From a cybersecurity perspective, the agreements were designed to evaluate frontier AI capabilities. Frontier AI refers to the most advanced models that could pose existential risks if misused. Pre-deployment evaluations are meant to catch problems before they emerge. The archived announcement noted that the agreements would help the government understand AI capabilities and the state of international AI competition, particularly against countries like China.
The timing of the page's removal is also noteworthy. It happened just a few days after the announcement, before most media outlets could fully report on it. The initial coverage was limited, but the missing page has now attracted attention from outlets like Reuters. This incident may become a case study in how government transparency in AI policy is handled.
For now, the only public record of the announcement is on the Wayback Machine. This underscores the importance of archiving government documents. Advocacy groups for government transparency have long urged agencies to maintain permanent public records of policy decisions. The CAISI missing page is a reminder that online government information can be ephemeral.
In conclusion, while no official explanation has been given, the disappearance of the CAISI page is a concerning development for those who believe that AI policy should be conducted in the open. The agreements with Google, xAI, and Microsoft represent a significant step in government-industry collaboration on AI safety. If the page was removed due to a simple technical error, the government should restore it immediately. If it was removed for strategic reasons, the public deserves an explanation. As AI continues to advance, trust in the institutions overseeing its development is paramount. This incident, though small, could have lasting implications for how the government communicates about AI regulation.
Source: Gizmodo News