Artificial intelligence policy continues to move from theory to implementation in Washington, with new defense contracting decisions, congressional scrutiny of federal data practices and fresh debates over how to power the next wave of data centers. This week’s developments span national security, grid reliability and online privacy, alongside a Senate Commerce hearing focused on practical AI applications and a key Environment and Public Works discussion touching nuclear oversight and chemical safety. Here’s a concise look at the policy shifts and oversight activity shaping the technology landscape this week.

In this week’s Nimitz Tech:

• Pentagon/OpenAI: OpenAI reached a Pentagon agreement allowing lawful uses of its AI systems while adding its own safety safeguards.

• Instagram: Instagram will alert parents to teens’ repeated self-harm searches, but only if families opt in to supervision tools.

• DHS Data: Democratic lawmakers are pressing major tech companies to disclose what user data they shared with DHS after reports that the agency used administrative subpoenas to identify social media accounts critical of ICE.

WHO’S HAVING EVENTS THIS WEEK?

Tuesday, March 3rd

• Senate Commerce, Science, and Transportation: “Less Hype, More Help: AI That Improves Safety, Productivity, and Care” at 10:15am. Watch here.

Wednesday, March 4th

• Senate Environment and Public Works: “Business meeting to consider the nomination of Douglas Weaver, of Maryland, to be a Member of the Nuclear Regulatory Commission; to be immediately followed by a hearing to examine a discussion draft of an original bill entitled, “the Toxic Substances Control Act Fee Reauthorization and Improvement Act of 2026.”“ at 10:00am. Watch here.

TECH NEWS DRIVING THE WEEK

In Washington

• The Trump administration announced that it would halt all government use of AI company Anthropic’s technology after the firm declined to permit certain applications, including domestic surveillance and autonomous weapons. Defense Secretary Pete Hegseth labeled the company a “supply chain risk to national security,” a designation Anthropic said it plans to challenge in court. The move could jeopardize billions of dollars in potential revenue and affect existing deployments of Anthropic’s AI assistant, Claude, which is integrated into multiple federal agencies, including the Pentagon. The decision has sparked debate within the tech industry over the appropriate boundaries for military AI use. Legal experts questioned the process behind the designation, suggesting the dispute may ultimately be resolved through litigation.

• OpenAI then announced it had reached an agreement with the Department of Defense to provide its artificial intelligence technology for classified systems, shortly after President Trump ordered federal agencies to stop using rival Anthropic’s AI tools. The Pentagon had required contractors to permit “all lawful uses” of their technology, a condition Anthropic declined to accept without additional safeguards against domestic surveillance and autonomous weapons, leading Defense Secretary Pete Hegseth to designate the company a national security risk. OpenAI, led by Sam Altman, agreed to the Pentagon’s lawful-use standard while negotiating the ability to implement technical guardrails and embed personnel to help oversee safety. The deal positions OpenAI to expand its defense business amid Anthropic’s exclusion and reflects broader competition among major AI firms, including Google and Elon Musk’s xAI, for military contracts.

• Democratic lawmakers on the House Committee on Homeland Security have asked major technology companies to disclose how much user data they have provided to the Department of Homeland Security following reports that the agency issued administrative subpoenas seeking identifying information about social media accounts that track or criticize Immigration and Customs Enforcement. Letters sent to companies including Apple, Amazon, Google, Meta, Microsoft, Reddit, Snap, TikTok and X request details on how many subpoenas were received, what information was sought, how often companies complied and whether users were notified. Administrative subpoenas do not require judicial approval and have historically been used in serious criminal investigations, but reporting found that DHS expanded their use last year to identify anonymous online critics. Some lawmakers have announced plans to introduce legislation to limit the agency’s authority to use such subpoenas.

National

• The PJM Interconnection, which operates the electric grid across much of the Mid-Atlantic and Midwest, has proposed new rules that could affect how large data centers connect to and pay for electricity. The changes would limit certain cost-saving arrangements for large facilities that generate their own power on-site, particularly for projects over 50 megawatts, a size common for major data centers. PJM says the reforms are needed to handle rapid growth in power demand while maintaining grid reliability and ensuring costs are fairly shared. However, industrial and energy groups warn that the proposal could make on-site power generation less financially attractive, potentially raising electricity costs for large facilities and increasing reliance on the broader grid. The outcome could influence how future data centers are built and powered, especially as companies look for reliable and affordable energy to support expanding AI operations.

International

• Instagram will begin notifying parents when their teenage children repeatedly search for content related to suicide or self-harm, provided both parent and teen are enrolled in the platform’s parental supervision tools. The alerts, which will be sent through email, text, WhatsApp and in-app notifications, aim to give parents information and access to expert resources to help guide conversations with their children. The feature will roll out in the United States, Britain, Australia and Canada next week, with other regions to follow later this year. Meta is introducing the change as it faces trials in two states over claims that its platforms are addictive and harmful to young users. While some experts say the alerts are a positive step, they caution that the opt-in requirement limits their reach and argue that broader changes to platform design and content moderation may be necessary to better protect teenagers.

• South Korea approved Google’s request to export detailed geographic map data, reversing a longstanding restriction that had limited Google Maps’ functionality in the country. The decision allows Google to move government-inspected map data from domestic servers to its global data centers, excluding sensitive military and security information. As a result, Google is expected to introduce full navigation features such as real-time driving and walking directions, which have been unavailable in South Korea, where users have relied primarily on domestic apps like Naver and Kakao. The policy shift comes amid trade tensions with the United States and pressure from the Trump administration over regulatory treatment of American tech firms. While some industry figures warn that Google’s expanded presence could threaten local companies, others say increased competition may improve services and boost tourism.

Just for Fun

Join a free, hour-long guided tour of the National Museum of Asian Art’s Japanese collection, featuring highlights such as paintings, prints, screens and ceramics. Tours are offered at noon and are open to the public.

ADD TO THE NIMITZ NETWORK

Know someone else who would enjoy our updates? Feel free to forward them this email and have them subscribe here.

Additionally, if you are interested in our original publication on veterans affairs policy, check it out below: