Public Policy & Governance

The White House is preparing an executive order targeting Anthropic even as the company's lawsuit challenging its Pentagon supply chain risk designation proceeds in federal court, according to Wired. The administration has refused to rule out further action against the AI startup, which was designated a supply chain risk after refusing to let the Department of Defense use its technology for domestic surveillance. Microsoft has filed a brief supporting Anthropic, arguing the First Amendment prohibits government coercion of private actors to rewrite code for government purposes, as reported by Financial Times.

Anthropic told the court it could lose billions of dollars in revenue this year and urged expedited proceedings, according to Bloomberg. The case represents the first major test of whether the government can compel AI companies to compromise their safety protocols for national security purposes. The Electronic Frontier Foundation filed an amicus brief arguing that forcing companies to modify AI systems for surveillance sets dangerous precedent, per EFF.

China restricted state-run enterprises and government agencies from running OpenClaw AI apps on office computers within days of the agentic AI's emergence, moving to defuse potential security risks as companies and consumers experimented with the technology, according to Bloomberg. The rapid administrative action contrasts sharply with Western democracies' slower legislative processes.

YouTube expanded its AI deepfake detection tool to a pilot group of politicians, government officials, and journalists, allowing them to flag unauthorised likenesses for removal, per TechCrunch and The Verge. The tool, already available to millions of content creators, represents platform self-regulation filling gaps left by absent comprehensive AI legislation. Meanwhile, the UK opened consultation on a scaled-back digital ID plan, attempting to win over a sceptical public by potentially excluding children and some personal data, according to Financial Times.

The family of a child critically injured in one of Canada's worst mass shootings filed suit against OpenAI, arguing the company knew the 18-year-old perpetrator was planning a mass casualty event based on his ChatGPT conversations describing violent scenarios involving guns but failed to contact authorities, according to The Guardian and BBC. The lawsuit comes days after OpenAI's CEO said he would apologise to families in the remote Canadian town.

The case raises fundamental questions about AI providers' duty to warn and the legal boundaries of content moderation. No established legal framework exists for when AI companies must report user conversations to authorities, creating tension between privacy protections and public safety obligations. The lawsuit will test whether AI providers face liability for harms they could have prevented through monitoring and reporting, similar to social media platforms' evolving duties regarding extremist content.

Meta's Oversight Board declared the company's methods for identifying deepfakes are not robust or comprehensive enough to handle how quickly misinformation spreads during armed conflicts like the Iran war, calling on Meta to overhaul how it surfaces and labels AI-generated content, according to The Verge and BBC. The board's recommendation comes as fake AI content about the Iran war floods X, with the platform's Grok AI failing to accurately verify video footage and sharing its own AI-generated images about the war, per Wired.

The criticism highlights implementation gaps in existing content authentication standards. While C2PA cryptographic signatures exist to verify content provenance, adoption remains limited and easily circumvented. The Oversight Board's intervention represents rare institutional pressure on Meta to strengthen AI content moderation specifically during armed conflicts, when misinformation carries highest stakes. Meanwhile, Iran has begun targeting Gulf data centers in bombing campaigns, bringing cyber-physical warfare directly into civilian infrastructure, according to The Guardian.