Public Policy & Governance

Anthropic's lawsuit challenging the Department of Defense's designation of it as a supply-chain risk has drawn amicus support from CDT and ACLU, framing the dispute as a First Amendment and due process issue rather than purely a procurement matter. According to CDT, the designation occurred in retaliation for Anthropic establishing red lines against DOD use of its systems for fully autonomous lethal operations. Lawfare's reporting indicates Judge Boasberg is considering the case alongside other challenges to executive branch procurement decisions, suggesting judicial scrutiny of how agencies wield supply-chain security designations as tools to compel vendor compliance with policy preferences.

The timing is significant: this follows the Pentagon's controversial agreement allowing OpenAI technology in classified environments, as reported by MIT Technology Review, creating a stark divide between AI vendors willing to support military applications without restrictions and those imposing ethical boundaries. The legal challenge could establish whether federal agencies can effectively blacklist companies for declining to participate in specific defense use cases, or whether such designations require concrete security justifications rather than policy disagreements.

Chancellor Rachel Reeves announced £1 billion in quantum computing funding over four years, explicitly framing the investment as a response to the UK's failure to retain AI leadership, according to The Guardian. Technology Secretary Liz Kendall stated the government will not allow quantum talent to slip away as happened with AI, where UK academic research failed to translate into domestic commercial dominance. Bloomberg notes governments increasingly view quantum computing as critical to national security, suggesting this funding reflects strategic competition concerns rather than pure R&D investment.

Reeves separately pledged the UK will achieve the fastest AI adoption rate in the G7, as reported by Bloomberg, positioning AI deployment as central to economic growth strategy. However, the government has not detailed specific regulatory changes or procurement reforms to accelerate enterprise adoption, leaving the pledge's implementation mechanisms unclear. The Financial Times reports the funding aims to prevent UK quantum startups from being acquired by overseas rivals, indicating the strategy includes capital controls or regulatory reviews of foreign investment in strategic tech sectors.

Senator Elizabeth Warren challenged the Pentagon's decision to grant xAI access to classified networks, citing Grok's creation of harmful outputs and potential national security risks, according to TechCrunch. The scrutiny intensified as three Tennessee teens filed a proposed class-action lawsuit against xAI alleging Grok generated sexualised images and videos of them as minors, as reported by The Verge. The lawsuit directly contradicts the Pentagon's apparent determination that xAI systems meet security and safety standards for classified use.

This creates a politically untenable situation: the Defense Department has elevated xAI to classified access while the same system faces allegations of generating child sexual abuse material, providing congressional critics with concrete evidence of inadequate safety testing. Warren's inquiry will likely focus on what vetting process the Pentagon applied and whether Musk's political proximity to the administration influenced the decision. The contrast with Anthropic's supply-chain risk designation for refusing autonomous weapons use becomes stark when DOD grants classified access to a system with documented safety failures.

Encyclopedia Britannica and Merriam-Webster filed suit against OpenAI alleging the company used nearly 100,000 copyrighted articles for training without permission and that GPT-4 'memorized' their content, producing substantially similar outputs, as reported by The Verge and TechCrunch. The memorisation allegation is legally significant because it undermines fair use defences that rely on transformative use arguments — if the model reproduces content verbatim or near-verbatim, transformation is difficult to demonstrate.

These plaintiffs differ from previous news publisher lawsuits because reference content is explicitly designed for accuracy and factual authority, making OpenAI's use particularly hard to justify as commentary or criticism. The publishers will likely argue their content provides the factual foundation that makes ChatGPT responses appear authoritative, directly substituting for the original works. This case could establish precedent for how courts treat training on reference materials versus news articles or creative works, potentially creating tiered copyright protections based on content type.