About a week ago, the Commerce Department’s Center for AI Standards and Innovation (CAISI) announced a significant step in AI oversight: new agreements with Google DeepMind, Microsoft, and xAI allowing the government to inspect unreleased AI models before public deployment. However, that official announcement has mysteriously disappeared from the CAISI website.
The original URL now leads to an error page or redirects to the main CAISI page on the Commerce Department website. Reuters first noticed the disappearance on Monday afternoon. Using the Wayback Machine, the original text can still be retrieved: “Today, the Center for AI Standards and Innovation (CAISI) at the Department of Commerce’s National Institute of Standards and Technology announced new agreements with Google DeepMind, Microsoft and xAI. Through these expanded industry collaborations, CAISI will conduct pre-deployment evaluations and targeted research to better assess frontier AI capabilities and advance the state of AI security.”
What the Missing Page Contained
The archived announcement emphasized that these agreements “support information-sharing” and “ensure a clear understanding in government of AI capabilities and the state of international AI competition.” It built on earlier partnerships with Anthropic and OpenAI, signed back in 2024, which had also been renegotiated to align with CAISI’s directives from the secretary of commerce and America’s AI Action Plan.
The disappearance raises questions about transparency and the government’s commitment to public accountability in AI governance. While the page may have been removed due to a technical error, the timing — just days after publication — fuels speculation about potential revisions or political sensitivities surrounding AI regulation.
Background on AI Pre-deployment Evaluation
The concept of pre-deployment evaluations, often called “red-teaming” or “safety testing,” has become central to AI policy debates. The Biden administration’s 2023 executive order on AI encouraged such testing, and subsequent actions by the Department of Commerce aimed to institutionalize it. CAISI was established within NIST to lead these efforts. The agreements with major AI developers represent a voluntary framework, distinct from mandatory regulation, but they signal a push for proactive oversight.
Anthropic and OpenAI were early participants in 2024, setting a precedent. The inclusion of Google DeepMind, Microsoft, and xAI expands the scope. Notably, xAI, founded by Elon Musk, had previously been critical of government regulation, so its participation marks a shift in the industry landscape.
The Significance of the Missing Page
When a government agency removes an announcement about AI safety inspections, it invites scrutiny. Critics argue that this undermines the credibility of the agreements and suggests internal disagreements or pressure from the tech industry. The White House and Commerce Department have not yet responded to requests for comment, leaving observers to speculate.
Some experts point out that the page might be undergoing revision. However, the fact that it vanished completely — rather than being marked as “updated” — is unusual. The Wayback Machine capture shows the original text in full, but the live site currently shows a 404 then redirects to the CAISI homepage. This happens at a time when the administration is finalizing America’s AI Action Plan, which is expected to outline regulatory priorities.
Furthermore, the missing page could be linked to ongoing debates about the scope of government access to proprietary AI models. Companies like Google and Microsoft have trade secrets to protect, and the terms of data sharing might be contentious. The cancellation of the public announcement could indicate that the agreements are being renegotiated behind closed doors.
The international dimension is also relevant. The U.S. is competing with China and the European Union in AI development and regulation. The EU’s AI Act imposes strict rules, while the U.S. has favored industry self-regulation. The disappearance of the CAISI page might be a small but telling sign of how fragile these voluntary agreements are.
The article from Mike Pearl, originally published on May 11, 2026, highlighted the potential embarrassment for the administration if the removal was intentional. The fact that the announcement was only a week old when it vanished suggests a hasty decision.
Industry Reactions and Implications
While no official statements have been made from the companies involved, industry analysts have noted that the deals were seen as a win-win: companies get a seal of approval before launch, and the government gains early insight into frontier capabilities. The disappearance could spook investors and regulators alike.
Gizmodo, the original publisher, has requested comment from the White House and Commerce Department but has not received a reply. The article noted that the page was missing as of Monday night, and the redirect persisted. The public now relies on archived copies to understand what was agreed upon.
Cybersecurity experts have also weighed in. Some argue that the removal without explanation is a breach of trust. Others suggest it may be a simple technical glitch. However, given the sensitive nature of AI oversight, even a glitch would be poorly timed.
The incident also reflects broader challenges in AI policy. The government is trying to balance innovation with safety, and transparency is key. When announcements disappear, it erodes confidence in the process.
Looking ahead, the Commerce Department will likely need to address this gap. Whether they re-post the original text or release a revised version remains to be seen. For now, the only evidence of the agreement lives in digital archives.
In summary, the missing page is more than a broken link. It is a symbol of the opaque nature of AI governance, where crucial information can vanish without notice. As frontier AI models become more powerful, the public deserves clarity on how the government intends to keep them safe. The story of the missing announcement is a reminder that transparency in AI policy is still a work in progress.
Source: Gizmodo News