Public Policy & Governance

The Trump Department of Justice filed a brief in federal court last week urging dismissal of the NAACP's April 2026 lawsuit against xAI and its subsidiary MZX Tech, which alleges the company deployed dozens of methane gas turbines in residential north Mississippi neighborhoods without adequate environmental permitting. The DOJ's intervention — framing the case as one warranting federal deference rather than civil litigation — is a confirmed legal filing, not a political statement, and carries direct weight in the proceeding. The Guardian

The intervention is significant on two levels. First, it demonstrates the administration's willingness to deploy DOJ resources to shield AI infrastructure development from environmental and civil rights challenges — a posture consistent with its broader deregulatory AI agenda. Second, it puts the federal government in direct opposition to a major civil rights organization in a case with clear environmental justice dimensions, in communities that are majority Black. For policy professionals tracking the governance of AI datacenter buildout, this case is a leading indicator of how the administration intends to handle infrastructure permitting conflicts: preemptively, and in favor of deployment speed over local regulatory compliance.

The Anthropic-Pentagon conflict has moved beyond political friction into active constitutional litigation territory. EFF and allied organizations have filed an amicus brief arguing that the Pentagon's actions against Anthropic — which Foreign Policy links to Anthropic's public stances on AI safety and its resistance to administration directives — violate the First Amendment because they are motivated by a desire to punish an uncooperative company rather than by legitimate national security concerns. This is a confirmed legal position in active proceedings, not a speculative claim. EFF

The case matters structurally because the administration has no comprehensive AI regulatory statute to enforce — it is operating through defense contracts, security clearances, and procurement leverage. Using those tools to discipline companies for their public policy positions is precisely the kind of unconstitutional coercion that EFF's brief targets. Separately, computer scientist Stuart Russell, writing in The Guardian, flags a different Anthropic development — early signs of recursive self-improvement in the company's models — as potentially more consequential than the political conflict, arguing that regulatory frameworks are not remotely prepared for that risk horizon. The Guardian

Senators Ted Cruz (R-TX) and Ron Wyden (D-OR) introduced the JAWBONE Act last week — Justice Against Weaponized Bureaucratic Overreach to Networked Expression — creating a federal private right of action against government officials who coerce broadcasters, interactive computer services, or AI providers into restricting lawful, First Amendment-protected speech. The bill also mandates a transparency regime for government communications with these intermediaries. This is introduced legislation, not enacted law, but the Cruz-Wyden pairing — ideologically opposed senators who rarely co-sponsor — signals this has genuine cross-aisle momentum. EFF

The explicit inclusion of AI providers in the bill's scope is new ground. Prior jawboning legislation focused on social media platforms; extending it to AI model developers reflects a recognition that AI companies are increasingly subject to the same governmental pressure dynamics that plagued social platforms during the COVID content moderation era. For compliance officers and government affairs teams at AI firms, the bill's transparency provisions — requiring disclosure of government communications — are operationally significant regardless of whether the cause of action is ever invoked.

A coalition of European civil society organizations, coordinated through AlgorithmWatch, has published a joint analysis of the recently approved AI Omnibus — the EU's package of amendments intended to simplify AI Act compliance obligations. The analysis concludes that the Omnibus rolls back fundamental safeguards, including provisions on high-risk AI systems and transparency requirements, before those provisions have even entered into force. This represents a confirmed legislative amendment to an enacted regulation, not a proposal. AlgorithmWatch

The process critique is as important as the substantive one: the organizations argue that using an omnibus simplification vehicle to weaken foundational rights protections without dedicated parliamentary scrutiny sets a dangerous precedent for future EU digital legislation. This mirrors a pattern seen in financial services regulation, where simplification mandates have historically been used to roll back post-crisis protections during periods of growth pressure. The political context — EU competitiveness concerns and pressure to match the US's permissive AI posture — is the driving force behind the Omnibus, and the industry lobbying that supported it has been effective precisely because it framed weakening as streamlining.

Two regional developments illustrate the widening divergence in national AI governance philosophies. In Latin America, Brazil is advancing a formal AI regulatory framework while Argentina under President Milei has explicitly rejected regulation, positioning itself as a permissive jurisdiction to attract AI investment. Foreign Policy The Brazil-Argentina split is not merely ideological — it will create concrete compliance asymmetries for multinational AI operators in Mercosur and complicate any future regional harmonization effort.

In Australia, Liberal MP Andrew Hastie has delivered a major address comparing AI to the Cold War nuclear arms race, warning that without dramatically scaled AI investment, Australia risks becoming a 'supplicant state' with strategic sovereignty constrained by US-China AI dominance. The Guardian Hastie's framing — from the opposition benches — is explicitly geopolitical rather than regulatory, but it reflects a growing consensus across Five Eyes governments that AI governance cannot be separated from defense and strategic autonomy questions. Australia currently lacks a comprehensive AI regulatory statute, and Hastie's speech adds pressure on the Albanese government to articulate a national AI strategy that goes beyond the existing voluntary frameworks.