About a week ago, the Commerce Department’s Center for AI Standards and Innovation (CAISI) announced a significant deal with AI companies Microsoft, xAI, and Google, allowing the government to inspect unreleased AI models before public release. The announcement was published on May 5, 2026, and detailed expanded industry collaborations for pre-deployment evaluations and targeted research. However, as of this week, that page has disappeared from the CAISI website, with the original URL now redirecting to the main CAISI landing page. Only a cached version via the Wayback Machine preserves the original text.
The Vanishing Announcement
The missing page contained a detailed statement from CAISI, explaining that the agreements build on previously announced partnerships and have been renegotiated to align with directives from the secretary of commerce and America’s AI Action Plan. The archived copy reads: "These agreements support information-sharing, ensuring a clear understanding in government of AI capabilities and the state of international AI competition." The deletion was first noticed by Reuters, who reported that the URL initially returned an error message before being redirected. The timing is peculiar given the growing scrutiny of AI regulation and the government's role in overseeing frontier models.
Details of the Agreements
The now-missing page outlined that CAISI would conduct pre-deployment evaluations of frontier AI models from Google DeepMind, Microsoft, and xAI. These evaluations are intended to assess capabilities and security risks before models are released to the public. Similar agreements had already been signed with Anthropic and OpenAI back in 2024, setting a precedent for government oversight. The expanded deals reflect a continued effort to keep pace with rapidly advancing AI technologies, particularly as competition with international players intensifies. The archived text also highlighted that the evaluations would cover both safety and competitive intelligence, helping the government understand where U.S. AI capabilities stand globally.
Broader Context of AI Oversight
The disappearance of the page occurs amid ongoing debates about the transparency and independence of AI regulatory bodies. CAISI was established to develop standards and conduct evaluations, but its operations have been subject to political pressures. The 2024 agreements with Anthropic and OpenAI were hailed as a milestone for proactive safety measures, but subsequent deals have faced criticism for lack of public disclosure. The removal of this announcement—whether accidental or intentional—raises further concerns about whether the public can trust that government oversight of AI is being conducted openly. Experts point out that such deletions can undermine the credibility of regulatory processes, especially when they involve major tech firms with significant lobbying power.
Possible Reasons for the Removal
There are several plausible explanations for why the page went missing. A technical glitch, where a URL was inadvertently changed or a page was unpublished during a website redesign, is one possibility. However, given the sensitivity of AI vetting, some suspect a deliberate removal due to internal policy shifts or pressure from involved companies. The redirect to the main CAISI page suggests a controlled action rather than a simple 404 error. Another theory is that the page contained information that was deemed too revealing about the extent of government access to proprietary models, potentially affecting trade secrets or competitive advantages. Without an official explanation from the Commerce Department or the White House, speculation continues.
Impact on AI Policy and Public Trust
The loss of this page has practical implications for researchers, journalists, and advocates who rely on government transparency to hold agencies accountable. The cached version remains a key source, but its impermanence highlights the fragility of digital records. If the government can quietly remove pages without notice, the public may question what other information has been deleted. This incident could erode trust in the AI evaluation process, especially if stakeholders suspect that companies are influencing the suppression of unfavorable findings. Moreover, it complicates ongoing efforts to establish international norms for AI safety, as other nations may perceive the United States as inconsistent in its regulatory commitments.
Industry and Expert Reactions
While the article awaits official comment, various AI policy experts have weighed in on social media and blogs. Many emphasize that pre-deployment evaluations are essential for catching dangerous capabilities before they reach millions of users. The removal of the announcement, regardless of intent, creates confusion about which models are currently under review and what specific criteria are being used. Some critics argue that the government should maintain a public registry of all active evaluation agreements to ensure accountability. Others note that the timing—just weeks before a major tech conference—could be coincidental but still raises eyebrows. The lack of a clear explanation leaves the door open for conspiracy theories, which further muddies public discourse on AI safety.
Historical Precedents of Government Page Removals
This is not the first time a government page has disappeared without explanation. During previous administrations, pages related to climate change, healthcare, and immigration were altered or removed during policy transitions. However, in the AI domain, where rapid development outpaces regulation, such removals are particularly damaging. The 2024 agreements with Anthropic and OpenAI were extensively documented and remain visible, but the newer deals with Google, xAI, and Microsoft were only briefly public. If the government is selectively deleting pages based on political or corporate pressure, it sets a dangerous precedent for future oversight. Watchdog groups have already called for a thorough investigation into the incident and for the immediate restoration of the page.
Technical and Legal Considerations
From a technical standpoint, the redirect indicates that a server-side change was made. It is unlikely to be a simple caching issue. The URL structure suggests that the page was intentionally unpublished or moved without a redirect to the new location. Legal experts note that while government websites are not bound by the same archiving rules as libraries, the removal of a public announcement could violate internal policies on record-keeping. The Presidential Records Act and other federal statutes may apply, especially if the announcement is considered an official record of agency action. However, enforcement of these laws is often weak, and the public may never learn the full story unless journalists continue to press.
As of this writing, the Commerce Department has not responded to requests for comment. The missing page remains redirected, and the only full copy available is via the Wayback Machine. This incident underscores the need for robust digital preservation of government materials, particularly in fast-moving fields like artificial intelligence. Until the page is restored or an explanation is provided, the AI community will continue to watch closely, and the question of what else might be missing from government websites will linger.
Source: Gizmodo News