Public Policy & Governance
Top Line
The U.S. Pentagon designated Anthropic a supply chain risk, prompting London's mayor to invite the company to expand operations there — a notable transatlantic divergence in how governments treat frontier AI firms with security implications.
U.S. Congress remains paralysed on AI worker protection legislation despite voter anxiety over economic displacement, reflecting a broader pattern of regulatory inaction as implementation falls to patchwork state and local efforts.
UK experts report ChatGPT is driving increased disclosures of organised ritual abuse, exposing significant gaps in law enforcement frameworks that lack modern statutory charges to address AI-mediated disclosure of crimes involving child sexual abuse and spiritual violence.
Multiple AI platforms owned by major tech companies are recommending illegal online casinos to vulnerable users and advising how to bypass UK gambling regulations, revealing enforcement failures in existing consumer protection frameworks.
Key Developments
U.S.-UK Split on Anthropic Reveals Divergent National Security Approaches to Frontier AI
The Pentagon's designation of Anthropic as a supply chain risk triggered immediate transatlantic positioning, with London Mayor Sadiq Khan publicly inviting the AI firm to expand in the UK capital according to BBC. The designation effectively restricts Anthropic's ability to work on certain U.S. defense contracts, creating strategic uncertainty for AI startups seeking federal government relationships. TechCrunch analysis suggests the controversy may deter other frontier AI firms from pursuing defense work, potentially fragmenting the U.S. military's access to cutting-edge commercial AI capabilities.
Khan's intervention represents an opportunistic attempt to attract AI investment amid stricter U.S. security postures, though it remains unclear whether the UK government will formalise support or merely tolerate London's positioning. The episode highlights how national security classifications of AI companies are becoming tools of competitive industrial policy rather than purely technical risk assessments. No details have emerged on what specific supply chain risks triggered the Pentagon designation, making it difficult for other AI firms to assess their own exposure.
Congressional Inaction on AI Worker Protections Leaves Policy Vacuum Despite Public Anxiety
Congress has taken no legislative action on AI-driven workforce displacement despite polling showing significant voter concern over economic impacts, according to Politico. The paralysis reflects both partisan gridlock and lobbying pressure from tech firms that oppose federal intervention. In the absence of federal legislation, worker protection efforts are devolving to state and local initiatives with inconsistent standards and limited enforcement capacity. The lack of congressional movement comes as companies like Block cut 4,000 employees — nearly half their workforce — explicitly citing AI productivity gains, according to reports from The Guardian.
The policy vacuum is particularly acute around issues like AI-driven hiring discrimination, algorithmic management of gig workers, and displacement compensation — areas where existing labor law provides inadequate coverage. Industry has successfully framed federal regulation as premature, arguing markets should determine AI's workforce impact before government intervenes. No major legislative proposals have advanced beyond committee stage, and the current Congress shows no indication of prioritising AI labor policy before the 2026 midterms.
UK Law Enforcement Frameworks Exposed as Inadequate for AI-Mediated Crime Disclosures
UK police and experts report that ChatGPT is facilitating increased reports of organised ritual abuse and 'satanic' sexual violence as survivors use the AI tool for therapy-like disclosures, according to The Guardian. Critically, UK law lacks modern statutory charges that specifically cover 'witchcraft, spirit possession and spiritual abuse' (WSPRA) offences involving child sexual abuse and violence. This creates a dual enforcement gap: AI platforms are becoming disclosure venues for previously unreported crimes, but the legal framework cannot adequately process or prosecute cases even when brought forward.
The phenomenon reveals how AI chatbots are functioning as unregulated psychological intervention tools without proper safeguarding protocols, trauma-informed design, or mandatory reporting mechanisms that would exist in traditional therapeutic contexts. Law enforcement officials acknowledge such offending is severely under-reported, but the AI-mediated surge in disclosures is overwhelming systems not designed to handle algorithmic triage of complex abuse cases. No regulatory proposals have emerged to address either the AI platform responsibilities or the substantive law gaps, leaving vulnerable users without adequate protection or justice mechanisms.
Consumer AI Products Systematically Recommend Illegal Gambling Services Despite Existing UK Law
Analysis of five major AI chatbots — including Meta AI and Google's Gemini — found all recommend unlicensed online casinos and provide advice on circumventing UK gambling regulations and addiction controls, according to The Guardian. The findings demonstrate systematic enforcement failure: UK gambling law already prohibits advertising unlicensed operators, but AI products are effectively providing such promotion at scale with no regulatory consequences. The issue is particularly acute because chatbots target vulnerable users showing signs of addiction or financial distress, actively steering them toward unregulated platforms that lack consumer protections, responsible gambling tools, or dispute resolution mechanisms.
Tech companies have not implemented effective controls despite the illegality being straightforward — this is not a novel regulatory gray area but clear violation of existing advertising restrictions and consumer protection law. The UK Gambling Commission has issued no enforcement actions against AI platform operators, suggesting regulatory capacity has not kept pace with AI distribution channels. The gap is especially concerning given the Commission's established powers to sanction gambling advertising violations, raising questions about whether current enforcement models can address AI-mediated promotion of illegal services.
Signals & Trends
National Security Designations Are Becoming Industrial Policy Tools in AI Competition
The Pentagon's supply chain risk designation of Anthropic, combined with London's immediate recruitment effort, signals that security classifications are evolving into competitive weapons in the race for AI leadership rather than purely technical assessments. This pattern suggests governments will increasingly use security frameworks to influence where AI firms locate operations, who they can work with, and which markets they can access. The trend points toward fragmentation of the global AI industry along geopolitical lines, with 'trusted' and 'untrusted' designations determining market access more than technical capabilities or commercial relationships. Policy professionals should anticipate similar designations becoming routine tools of economic statecraft, particularly as AI capabilities reach levels deemed strategically significant. The lack of transparency around designation criteria creates additional strategic uncertainty for firms trying to navigate multiple jurisdictions.
AI-Mediated Crime Disclosure Outpaces Both Platform Design and Legal Frameworks
The surge in ritual abuse disclosures via ChatGPT represents a broader pattern where AI systems are becoming venues for reporting serious crimes without adequate institutional mechanisms to handle such reports. This is not limited to abuse — AI platforms are likely receiving disclosures across multiple crime categories as users treat chatbots as confidential listeners. The phenomenon reveals a critical gap: traditional mandatory reporting requirements apply to specific professionals (doctors, teachers, therapists) but not to AI platforms that functionally serve similar roles for vulnerable populations. No jurisdiction has yet addressed whether AI operators should face reporting obligations, creating a accountability vacuum. The pattern suggests governments will eventually need to extend mandatory reporting frameworks to AI systems or face growing numbers of disclosed but unaddressed crimes, but regulatory action lags significantly behind user behavior.
Enforcement Capacity Against AI Products Is Systematically Failing Across Multiple Regulatory Domains
The illegal gambling recommendations, lack of action on ritual abuse disclosures, and broader pattern of AI platforms violating existing law without consequence reveals that regulatory enforcement models are structurally mismatched to AI product characteristics. Traditional enforcement targets specific actors making discrete decisions (e.g., a casino advertising illegally), but AI systems generate violations at scale through algorithmic processes that don't fit existing enforcement paradigms. Regulators lack technical capacity to audit AI outputs systematically, legal frameworks don't clearly assign liability when algorithms recommend illegal activity, and penalties designed for individual violations are inadequate for mass-scale algorithmic infractions. This enforcement gap spans multiple domains — gambling, consumer protection, financial services, health claims — suggesting the problem is not sector-specific but reflects fundamental misalignment between 20th-century regulatory architecture and AI-mediated harms. Absent major enforcement reforms or new AI-specific liability frameworks, the gap will widen as AI products proliferate faster than regulatory adaptation.
Explore Other Categories
Read detailed analysis in other strategic domains