7-minute read | 1,500 words
What to know this week
Pentagon reaches new deal with AI providers.
The Pentagon announced it has reached agreements with several artificial intelligence (AI) companies to deploy advanced capabilities in the Department of Defense’s (DoD) networks.
White House considering vetting AI models before public release.
The Trump administration is considering moving away from its non-involved AI approach to now imposing oversight requirements on AI models before their public releases.
This week's full stories
Pentagon reaches deals with seven AI providers.
THE NEWS
On Friday, the Pentagon announced that it had reached agreements with seven AI providers. With these various agreements, the Pentagon is looking to integrate advanced AI capabilities into the DoD’s classified networks.
The seven AI companies include SpaceX, OpenAI, Google, NVIDIA, Reflection, Microsoft, and Amazon Web Services. Notably, several of these companies were already working with the administration.
During the announcement, the Pentagon stated:
“These agreements accelerate the transformation toward establishing the United States military as an AI-first fighting force.”
Notably, the Pentagon did not elaborate on how these new AI tools would be used or implemented.
The Pentagon stated:
“Access to a diverse suite of AI capabilities from across the resilient American technology stack will give war fighters the tools they need to act with confidence and safeguard the nation against any threat.”
THE KNOWLEDGE
With these new agreements, the Pentagon is looking to avoid “vendor lock”. Alongside avoiding any vendor lock issues, the Pentagon officials are likely hoping that the new agreements will encourage Anthropic to drop its reservations about any broad lawful use of AI.
Though relations between the Trump administration and Anthropic appear to be improving, it is unclear how much tensions have calmed. For context, Anthropic and the Pentagon had a falling out earlier this year over concerns about how the government would be utilizing Anthropic’s systems. This eventually culminated in the Pentagon designating the AI company as a supply chain risk and President Trump ordering agencies to begin removing these services.
Afterwards, Anthropic filed lawsuits against the administration, which are still ongoing. However, relations have begun to cool in recent weeks after Anthropic released its new cybersecurity-centered AI model, Claude Mythos Preview. The White House has reportedly begun to develop new guidance that would walk back its previous order. Though still unreleased, this new guidance would allow federal agencies to continue using Anthropic’s services despite the supply chain risk designation.
As these new AI deals take shape, the relationship between Anthropic and the Trump administration will be critical.
THE IMPACT
Though the details of these contracts are unclear, the agreements signal that the Pentagon is moving aggressively to normalize AI inside classified environments. If implemented at scale, these new deals would continue the government’s efforts towards having AI become further embedded into core military decision-making, intelligence workflows, and cyber operations.
More importantly, the Pentagon’s decision to spread contracts across seven providers shows that the DoD is intentionally avoiding any dependency on any single company’s services as it had previously done. This multi-vendor approach would give the Pentagon greater leverage being able to choose between multiple providers and avoid being trapped by proprietary tools or licensing agreements.
As these deals take shape, it is likely that they will reignite the debate regarding how the military utilizes these AI services and what will be considered lawful usage.
Trump administration considering pivot on AI oversight requirements.
THE NEWS
On Monday, reports emerged that the Trump administration is looking to pivot away from its hands-off AI policies toward a more formal oversight regime. More specifically, the administration is debating whether to implement a new Executive Order (EO), which would establish a new AI working group. In this group, tech executives and federal officials would meet to examine oversight procedures alongside discussing plans to establish a formal government review process.
Additionally, this new working group would likely consider a variety of different oversight approaches with a review process potentially mimicking the one found in Britain.
Previously, the Trump administration had often focused on a strong deregulation approach when it came to managing AI. However, with new AI leadership in the White House, this could signal a potential shift in the administration’s overall AI policy.
Dean Ball, a senior AI advisor for the Trump administration in 2025, commented on the development noted:
“The technology is moving extremely fast, and there are few formal procedures, but they also don’t want to overregulate. It’s a tricky balance.”
THE KNOWLEDGE
In March 2026, David Sacks, the Trump administration’s AI czar, left his role. At the White House, Sacks was a key driver for the administration's AI deregulation efforts. Under Sack’s guidance, he advocated that the best way for the United States to win the global AI race was to move fast and cut any disruptive regulation. Further, Sacks often argued that any of the problems created by this deregulated AI approach, would be solved by the solutions created by AI.
This mentality has been a guiding principle for the Trump administration, which has attempted several times to impose a federal AI moratorium. Under the proposed moratoriums, which were originally introduced in May 2025, the Trump administration was looking to invalidate a vast majority of state AI laws.
The most recent attempt to impose this moratorium on state AI came through President Trump’s December 2025 EO “Ensuring a National Policy Framework for Artificial Intelligence.” In the order, the President ordered the creation of a new AI litigation task force charged with challenging state laws that the administration has deemed to interfere with interstate commerce alongside restricting funding for states that enforce restrictive AI regulations.
However, Sacks’s departure could mark a shifting point in the administration’s goals. Susie Wiles, the White House’s chief of staff, and Scott Bessent, the Treasury Secretary, have taken over Sacks’s previous responsibilities. Already, the two met with Anthropic after the company released Claude Mythos and have signaled their intent to have a bigger role in shaping AI policy.
That outreach suggests the White House may now view frontier AI model releases as something that needs more formal attention.
THE IMPACT
If the Trump administration moves forward with a formal AI model review process, it would represent a meaningful pivot away from its earlier deregulatory stance. Even if the group lacks major authority, it would still function as an informal gatekeeper for companies looking to release new models.
More broadly speaking, the shift reflects a growing concern that the risks of advanced AI systems are now tied to national security, cyber operations, and economic stability. If the Trump administration adjusts how it manages AI providers, it could shape lab development timelines, innovation potential, and how the federal government influences the sector as a whole.
This Week's Caveat Podcast: Trump’s battle on AI policy controls.
Dave Bittner and Ben Yelin look at how recent executive actions related to AI have begun to realign the relationship between federal and state policy. Additionally, the two look at a lawsuit involving a college student suing a dating application for reappropriating her image. Lastly, Ben sits down with Jen Sovada, Claroty’s Public Sector GM, to discuss national cybersecurity strategies and how these initiatives depend on public-private trust.
OTHER NOTEWORTHY STORIES
New Mexico seeks platform changes in Meta youth harm trial.
What: New Mexico is suing Meta looking for a judge to force platform changes.
Why: On Monday, New Mexico’s Attorney General Raul Torrez began his lawsuit against Meta. In this case, Torrez alleges that Meta intentionally designed its platforms to addict young users and failed to implement proper protections to prevent sexual exploitation of minors.
During this trial, the judge will decide whether these platforms have created a “public nuisance” under the state’s law. If the judge finds them to be in violation of the law, he could order a wide-range of remedies.
Meta pushed back on these allegations in their court filings stating that there is no scientific evidence to support the claims that social media has caused mental health problems and that many of the proposed changes would be impossible to implement.
A Meta spokesperson commented on the matter stating:
“The New Mexico Attorney General’s focus on a single platform is a misguided strategy that ignores the hundreds of other apps teens use daily.”
MAY 4, 2026 | Source: Reuters
Doctors’ social security numbers exposed by National Provider Directory.
What: The National Provider Directory inadvertently exposed the Social Security numbers (SSNs) of over one hundred health providers.
Why: On Friday, the Centers for Medicare and Medicaid Services (CMS) removed its National Provider Directory after the CMS admitted that a small number of SSNs were exposed in downloadable data.
Currently, at least 102 providers have had their full SSNs included in the database. While this database is primarily intended for insurers to use to create search tools for patients, the data was publicly accessible.
In a statement, the agency emphasized:
“The agency has taken steps to address it promptly and reinforce safeguards around data submission and validation.”
MAY 1, 2026 | Source: Politico