About a week ago, the U.S. Department of Commerce's Center for AI Standards and Innovation (CAISI) announced a significant step in AI governance: agreements with three major AI companies—Google DeepMind, Microsoft, and xAI—that would allow federal inspectors to examine unreleased AI models before they reach the public. The move was widely seen as a critical component of the government's AI Action Plan under the current administration. But as of Monday evening, the official announcement page has vanished from the CAISI website, replaced by a redirect to the center's main page. The only remaining record is an archived copy on the Wayback Machine, preserved by independent archivers.
The missing page originally carried the headline: “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.” It went on to explain that these agreements would enable CAISI to conduct pre-deployment evaluations and targeted research aimed at better assessing frontier AI capabilities and advancing the state of AI security. The announcement further noted that the partnerships had been “renegotiated to reflect CAISI’s directives from the secretary of commerce and America’s AI Action Plan.” Another key sentence from the archived document read: “These agreements support information-sharing, ensuring a clear understanding in government of AI capabilities and the state of international AI competition.”
The disappearance was first flagged by Reuters, which reported that the original URL initially yielded a “Sorry, we cannot find that page” error before being changed to a redirect. As of this writing, attempts to access the original link land on the CAISI main page with no explanation. This is not the first time a government announcement about AI regulation has been quietly taken offline, but the timing is particularly noteworthy given the ongoing debate over the pace and transparency of AI safety measures in the United States.
Background of the CAISI Agreements
CAISI was established within NIST to centralize the federal government's efforts in AI standards, testing, and innovation. Its mandate includes developing voluntary guidelines, conducting evaluations, and fostering collaboration with private sector leaders. The agreements signed with Google, xAI, and Microsoft are part of a broader framework that began in 2024 when Anthropic and OpenAI signed similar pre-deployment pacts. Those earlier deals set a precedent for government access to advanced AI models before public release, though many critics argued that they lacked enforcement mechanisms and transparency.
Under the terms of the expanded agreements, CAISI would gain privileged access to model architectures, training data descriptions, and evaluation results before launch. The stated goal was to identify potential risks—such as biases, security vulnerabilities, or capabilities that could be misused—and to inform federal policy directions. The archived page emphasized that these agreements were voluntary but aligned with the secretary of commerce's directives and the America's AI Action Plan, a strategy document released earlier in 2026 that called for robust safety testing without stifling innovation.
The companies involved—Google DeepMind, Microsoft, and xAI—represent some of the most advanced AI research organizations in the world. Google DeepMind is known for breakthroughs in protein folding (AlphaFold) and game-playing AIs. Microsoft has invested heavily in OpenAI and integrated AI into its product ecosystem, while xAI, founded by Elon Musk, focuses on accelerating AI development with an emphasis on scientific discovery. The combination of these three firms under a single federal oversight initiative signaled a broad industry willingness to cooperate—at least publicly.
Implications of the Missing Page
The removal of the announcement page raises several troubling questions. First, it undermines the principle of transparency that the government has repeatedly pledged to uphold regarding AI regulation. If the public cannot even verify the existence of such agreements, trust in the regulatory process erodes. Second, the timing—just days after the announcement and coinciding with the start of a new workweek—suggests either an internal mistake or a deliberate decision to remove information. The Commerce Department has not issued an official explanation for the disappearance.
Some analysts speculate that the page might have been taken down due to factual errors or incomplete disclosure. Others point to potential internal disagreements over the scope of the agreements or the level of detail shared with the public. It is also possible that the administration is reconsidering its approach to AI vetting under pressure from industry lobbyists who prefer less government scrutiny. Without official comment, the reason remains unclear.
The incident also highlights the fragility of digital government records. While the Wayback Machine preserves a snapshot, reliance on third-party archives is not a sustainable substitute for official transparency. In an era where AI capabilities evolve rapidly, the public deserves consistent and accessible information about how these systems are being evaluated for safety and alignment with national interests.
Reactions and Ongoing Debate
News of the missing page has sparked discussions among AI policy watchers, journalists, and advocacy groups. Some have drawn parallels to earlier instances where government notices about technology agreements were quietly altered or removed. The Coalition for Responsible AI, a watchdog group, issued a statement calling for the immediate restoration of the page and a full explanation. “The American people have a right to know how their government is overseeing the most transformative technology of our time,” the group said. “Silently deleting announcements does not inspire confidence.”
On social media, experts have contrasted the U.S. approach with that of the European Union, where the AI Act establishes a more codified system of transparency and risk classification. Others have noted that the missing page might simply be a technical glitch—a broken link or a site migration error. However, the lack of a prompt correction or alternative source has amplified skepticism.
The broader context includes a growing partisan divide over AI regulation. Some lawmakers advocate for a laissez-faire approach to maintain U.S. leadership, while others call for stronger guardrails to prevent harm. The disappearance of a key official page could be seen as a small but significant signal in that battle. It remains to be seen whether the information will be restored, whether a new version will appear with modifications, or whether the government will remain silent on the matter.
Additional Background on Federal AI Oversight
The CAISI agreements are part of a long history of federal involvement in technology safety. Earlier administrations established similar frameworks for emerging technologies such as nuclear energy, aviation, and pharmaceuticals. In the AI domain, the White House issued an executive order in 2023 that required developers of certain large models to share safety test results with the government. That order led to voluntary commitments from major AI firms. The 2026 AI Action Plan aimed to codify some of these practices into permanent agency procedures.
However, the voluntary nature of the current agreements means that companies can withdraw or renegotiate terms without public notice. The archived page emphasized that the agreements were “renegotiated to reflect CAISI’s directives,” implying that earlier versions existed but were not necessarily disclosed. This opacity fuels concerns that the public is being kept in the dark about the true extent of government-industry relationships in AI.
Another layer of complexity is the international dimension. The U.S. competes with China and other nations in AI development, and the government has a delicate balancing act: promoting innovation while mitigating risks. Public announcements of vetting deals can serve as a deterrent to reckless behavior by signaling that the government is monitoring the industry. Removing such announcements could be seen as an invitation for less responsible actors to accelerate their timelines without restraint.
The disappearance of the CAISI page also echoes similar incidents in other government IT systems. Federal websites have occasionally suffered from link rot, outdated content, or abrupt removals due to leadership changes. But in the context of AI regulation—where each day brings new breakthroughs and potential hazards—the loss of even a single announcement undermines the credibility of the oversight process.
As of press time, the Commerce Department and the White House had not responded to requests for comment. The archived page remains accessible via the Wayback Machine, offering a glimpse of what was officially stated just a week ago. Until the page is restored or an official explanation is provided, the public and policymakers alike are left to wonder whether the missing page is a minor administrative error or a symptom of deeper dysfunction in the government's approach to AI governance. The incident serves as a stark reminder that in the digital age, information can vanish as quickly as it appears, and that transparency requires constant vigilance.
Source: Gizmodo News