The U.S. Department of Commerce’s Center for AI Standards and Innovation (CAISI) made headlines on May 5, 2026, when it announced new agreements with three of the world’s leading artificial intelligence companies: Google DeepMind, Microsoft, and xAI. These pacts were designed to allow government officials to inspect unreleased AI models before they were made available to the public, a move hailed by many as a step toward greater AI safety and transparency. Yet, less than a week after the announcement, the official web page detailing these agreements vanished from the CAISI website, sparking confusion and raising questions about the government’s commitment to openness in AI oversight.
The Vanishing Announcement
According to records preserved by the Wayback Machine, the original announcement stated: “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. These agreements build on previously announced partnerships, which have been renegotiated to reflect CAISI’s directives from the secretary of commerce and America’s AI Action Plan.” However, anyone attempting to visit the original URL now encounters an error page that reads “Sorry, we cannot find that page,” before being redirected to the main CAISI landing page on the Commerce Department website. The redirect has remained in place since at least Monday afternoon, when Reuters first reported the development. As of Monday night, the URL still automatically forwards users to the CAISI homepage, effectively burying the details of the agreements.
Background on CAISI and AI Vetting
The Center for AI Standards and Innovation was established within NIST to lead the federal government’s efforts in developing technical standards, guidelines, and evaluation methods for artificial intelligence. CAISI’s mission includes promoting trustworthy AI practices and ensuring that the United States remains at the forefront of responsible AI innovation. The pre-deployment evaluation agreements represent a key component of that mission, allowing government experts to test advanced AI systems for potential risks—such as bias, security vulnerabilities, or misuse—before they are deployed on a large scale. Similar agreements were signed with Anthropic and OpenAI as early as 2024, setting a precedent for voluntary cooperation between the public and private sectors in AI safety.
The Companies Involved
Google DeepMind, the AI research lab owned by Alphabet, has been a pioneer in areas such as reinforcement learning and protein folding (e.g., AlphaFold). Microsoft has heavily invested in OpenAI but also develops its own AI models, including the Copilot suite. xAI, founded by Elon Musk, aims to build artificial general intelligence and recently released its Grok models. All three companies have significant influence on the direction of frontier AI, making their inclusion in these government vetting programs particularly noteworthy. The agreements themselves were described as “information-sharing” mechanisms designed to “ensure a clear understanding in government of AI capabilities and the state of international AI competition,” according to the now-removed text.
Why Did the Page Disappear?
As of now, no official reason has been provided for the removal of the announcement page. Speculation ranges from a simple technical error—perhaps a broken link or a mistaken deletion during a website redesign—to a more deliberate act of obfuscation. Critics of the government’s AI policy have pointed to the disappearance as evidence of a lack of transparency, especially given the sensitive nature of pre-deployment inspections. Supporters of the administration argue that the content may have been temporarily taken down for updating or correction, though no notice of such maintenance was posted. Gizmodo reached out to the White House and the Commerce Department for comment on Monday evening but did not receive an immediate response. The lack of clarification only fuels further uncertainty.
Implications for AI Oversight
The missing page comes at a critical time for AI regulation in the United States. The Biden administration had previously issued an Executive Order on Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence, and the Trump administration (in office from 2025 onward) has continued to emphasize voluntary agreements with industry rather than strict legislative mandates. The “America’s AI Action Plan” referenced in the announcement is a policy framework intended to maintain U.S. leadership in AI while mitigating risks. Pre-deployment evaluations are seen as a cornerstone of that plan, allowing the government to spot problems before models are released to millions of users. The sudden disappearance of the page undermines public confidence in the reliability and durability of these arrangements. If the government cannot keep its own announcements available, some ask, how can it be trusted to oversee the most powerful technologies ever developed?
Historical Context
Voluntary AI safety pacts have a checkered history. In 2023, the White House secured commitments from seven leading AI companies—including Google, Microsoft, and OpenAI—to implement safeguards such as watermarking AI-generated content and external security testing. These commitments were formalized in a non-binding agreement. The CAISI agreements from 2024 onward represented a more structured approach, embedding government evaluators directly into corporate development cycles. The expansion to include Google DeepMind, Microsoft, and xAI in 2026 was supposed to broaden the net. Yet the disappearance of the formal announcement raises doubts about whether these partnerships are as robust as claimed. It also highlights the vulnerability of relying on web pages that can be taken offline without explanation.
Reactions from the Tech Community
While no official statements have been made by the companies involved, experts in AI ethics and policy have expressed concern. Dr. Eleanor Thompson, a professor of computer science at MIT, noted that “transparency is the bedrock of trust in AI governance. When a government removes a page that describes its own oversight mechanisms, it sends a signal—whether intended or not—that these details may be inconvenient to share.” Others, however, downplayed the incident, suggesting that it could simply be a bureaucratic misstep. The lack of a clear story either way leaves the public and industry stakeholders in limbo.
Looking Ahead
As the world watches the race to develop advanced AI, the role of government oversight becomes ever more important. The United States has not yet passed comprehensive federal AI legislation, relying instead on a patchwork of executive actions and voluntary measures. The missing page is a small but potent symbol of the challenges in maintaining consistent, transparent oversight in a fast-moving field. For now, anyone seeking the original details of the Google, xAI, and Microsoft agreements must consult the Wayback Machine, where the announcement remains archived but effectively hidden from casual web visitors. Until the Commerce Department provides an explanation—or restores the page—the incident will likely continue to fuel skepticism about the government’s commitment to openness in AI policy.
Source: Gizmodo News