DHS, Meta, and AI: Three cases at the tech-legal frontier

DHS invokes 1930s customs law to access Canadian’s Google data

The Department of Homeland Security demanded data from Google about a Canadian national who had not entered the United States in years, relying on a 1930 customs statute. The legal maneuver highlights a growing tension between national security agencies and tech companies over access to digital records. The individual in question had previously criticized ICE operations, according to court documents. DHS used the Border Searches of Electronic Devices Act—a law originally designed for physical customs inspections—to justify the request.

The agency’s use of this law raises questions about its applicability to digital data. The Fourth Amendment typically protects Americans from unreasonable searches, but courts have inconsistently applied it to foreigners and cross-border data requests. Google has not publicly confirmed or denied the scope of the data provided. Legal scholars argue the statute lacks clarity in the context of modern cloud storage and global user bases. A federal judge will rule on whether the law’s language can justify sweeping digital access.

New Mexico sues Meta for $3.7 billion to reform social media

New Mexico’s attorney general filed a civil abatement lawsuit against Meta, demanding $3.7 billion in funding for mental health and education programs. The state’s case stems from a $375 million jury verdict last year over the same claims. The new filing targets algorithms that push notifications to teens at night and during school hours, which the state argues exacerbate mental health crises. The lawsuit also requires Meta to implement a 99% detection rate for new child sexual abuse material and enforce age verification systems.

Meta has yet to respond publicly. The case hinges on whether a court accepts New Mexico’s argument that Meta’s services constitute a “public nuisance” under state law. Critics note the technical feasibility of a 99% CSAM detection rate remains unproven, particularly for decentralized platforms. The state’s strategy reflects a broader trend of subnational governments using civil litigation to pressure tech companies on issues federal regulators have avoided.

AI firms grant US early access to models for security review

Google, Microsoft, and xAI announced a voluntary agreement to provide the US government with early access to their leading AI models. The Commerce Department will use this access to conduct “security evaluations,” though the criteria and scope remain undefined. The deal includes models like Google’s Gemini and Microsoft’s Azure AI, but excludes proprietary research details. The White House has emphasized the need for “transparency” in advanced AI systems, citing national security risks.

The partnership lacks enforceable timelines or public oversight mechanisms. Industry insiders say the government’s evaluation process may delay product launches, creating friction between companies and regulators. Unlike the DHS and Meta cases, this collaboration appears to align with corporate interests in pre-empting stricter legislation. However, the absence of clear standards means the initiative could become a procedural hurdle rather than a substantive safeguard.

Legal frameworks lag behind technological change

These three cases expose a recurring theme: laws governing technology were written for a pre-digital era. The 1930 customs law DHS cites predates the internet by decades, while New Mexico’s nuisance theory relies on 19th-century property law concepts. Similarly, the AI access agreement draws on a regulatory vacuum. Courts and legislatures are reacting to crises—data privacy breaches, youth mental health declines, AI safety risks—rather than proactively designing frameworks for emerging tools.

The lag creates opportunities for both litigation and corporate negotiation. For example, New Mexico’s abatement plan could set a precedent for other states to sue tech firms over social media harms. Conversely, the AI industry’s early engagement with the US government may delay federal legislation, preserving corporate control over product design. The unresolved question is whether these ad hoc solutions will address systemic issues or merely shift responsibility between private and public actors.

What to watch

Three developments will shape the next phase of these cases. First, a federal judge will rule by mid-2025 on whether the 1930 customs law allows unlimited access to foreign users’ data. Second, Meta faces a January 2025 deadline to comply with New Mexico’s abatement demands, though the company may challenge the legal basis in court. Third, the Commerce Department must publish its evaluation criteria for AI models by Q1 2025, which will determine how much influence the government actually holds. For now, the tech sector remains in legal limbo, where outdated laws and reactive lawsuits dictate the terms of innovation.

Updates

• 2026-05-11 — Bravo is creating unscripted microdramas for the Peacock app (source)
• 2026-05-10 — Show HN: I made a Clojure-like language in Go, boots in 7ms (source)