Policy & Regulation

Governor Greg Abbott announced that all new Texas data-centre projects must be audited by the Public Utility Commission and grid operator ERCOT - a sharp turn for a state whose loose regulation and cheap power made it second only to Virginia for data centres. The trigger is a staggering queue: ERCOT's interconnection requests doubled from 233GW in January to 474GW, about 90% data centres, more than five times the grid's all-time peak demand. Audits will demand power and water use, noise mitigation, light controls, tax-incentive use, and ownership details - after a voluntary survey that most operators simply ignored.

An AWS GraphRAG deployment in pharmaceutical research cut R&D cycles by 87% - initial discovery that took six months now closes in three weeks - by fusing siloed internal databases and public literature into one queryable knowledge graph on Amazon Neptune Analytics and Bedrock (running Claude). Every answer comes with verifiable citations and a mapped reasoning path, which is exactly what regulated industries need for compliance. The architecture is modular and, crucially, transferable: any enterprise drowning in fragmented legacy data can copy this pattern.

In a marked shift, the Bank of England is reviewing whether its rulebook can govern agentic AI in payments, trading, and operations - Deputy Governor Sarah Breeden told the ECB Forum in Sintra that frameworks were never built for autonomous agents and that keeping a human in the loop for every action is unrealistic. On the table: market-wide circuit breakers or kill switches, enhanced recovery letting one bank take over another's core functions, and joint BIS/Bundesbank simulations of AI herding risk - the danger that similar agents sell in unison under stress. With 52% of finance firms already adopting agentic AI (Cambridge CCAF) and the FSB consulting on 12 sound practices, compliance roadmaps just gained a chapter.

As the US export ban on Anthropic's Mythos and Fable 5 entered its third week, two Asian companies shipped self-declared alternatives: Beijing cybersecurity giant 360 unveiled Tulongfeng (automated vulnerability discovery) and Yitianzhen (automated cyber defence), with founder Zhou Hongyi calling vulnerability-finding AI a national strategic asset - while Tokyo's Sakana AI launched Fugu, an orchestration model it says stands shoulder-to-shoulder with Fable 5 and markets as frontier capability without export-control risk. The stakes for Anthropic are large: run-rate revenue crossed $47 billion in May 2026, and local rivals tuned to regional languages are moving in. The ban has since lifted - the competitors it spawned have not.

The EU has published a voluntary Code of Practice for marking and labelling AI-generated content, a practical playbook for meeting transparency duties that become law on August 2, 2026 under Article 50 of the AI Act. From that date, deepfakes and AI-generated text on public-interest matters must be labelled, and users must be told when they're talking to an AI. The code splits the work - model builders mark output in machine-readable form, deployers handle visible labelling - and companies have under two months to prepare.

Four days after Apple confirmed Siri AI won't launch in China, Huawei unveiled HarmonyOS 7, restructuring the OS around an agent-native architecture it calls the beginning of the agent era. Its assistant Xiaoyi, rebuilt as a system-level agent, now drives 2,100+ system capabilities and coordinates 2,000+ third-party AI agents, atop the upgraded openPangu foundation model. With HarmonyOS already past iOS in China's smartphone share, independence forced by US sanctions has become a structural advantage in the one market Apple can't reach at the AI level.

A US export-control directive forced Anthropic to abruptly disable Fable 5 and Mythos 5 for all foreign nationals on June 13, just four days after launch - briefly cutting off even its own overseas staff. Washington cited a jailbreak vulnerability; Anthropic disputed its severity but had to pull global access because it couldn't filter users by nationality in real time. Europe and Canada reacted with alarm, treating it as proof that frontier-AI access can be switched off by a single government overnight.

Anthropic and Tata Consultancy Services (TCS) struck a partnership that puts Claude in the hands of 50,000 TCS employees across 56 countries and builds Claude-powered products for banks, insurers, healthcare, and the public sector. As a self-described "customer zero," TCS is deploying Claude across its own engineering, finance, legal, and sales teams while packaging it into industry offerings like claims processing and lending advisory. The deal also deepens Anthropic's reach in India, now its second-largest market.

The US government issued an export-control directive forcing Anthropic to immediately disable Fable 5 and Mythos 5 for all customers, citing national security and a reported jailbreak. Anthropic is complying but disputes the basis, arguing the cited technique surfaces only minor, already-known vulnerabilities that rival models can find without any bypass. Every other Claude model remains unaffected and available.

Anthropic signed a multi-year global alliance with IT-services giant DXC Technology to weave Claude into the core systems it operates for the world's largest banks, airlines, insurers, and government agencies. DXC will train tens of thousands of Claude-certified forward-deployed engineers and has already used Claude to write 95%+ of the code for its new AI-native managed-services platform, DXC OASIS. The rollout begins in insurance, legacy-code modernization, cybersecurity, and application services.

Anthropic published two policy proposals arguing that AI is outpacing a policymaking process built for a slower era. Its Advanced AI Framework would give government legal authority to block or deter dangerous model deployments, backed by transparency, independent evaluation, and revenue-scaled penalties - aimed only at the largest frontier developers (models above 10^25 FLOPs). A companion Economic Policy Framework focuses on preparing workers and sharing AI's financial gains.

Anthropic analyzed 832 accounts banned for malicious cyber activity (March 2025-March 2026), mapping them to the MITRE ATT&CK framework. The data shows attackers increasingly using AI in later, more complex attack stages, with the share rated medium-risk-or-higher jumping from 33% to 56% across the year. Anthropic argues the traditional ways of gauging an attacker's threat level no longer hold, and is in talks with MITRE about updating the framework for agentic, AI-driven attacks.

The U.S. FTC is reportedly increasing scrutiny of Microsoft's AI and cloud practices, including outreach to rivals. For platform teams, this signals that cloud bundling, model partnerships, and distribution advantages may become central to AI antitrust narrativesnot just consumer harm arguments.

OpenAI told U.S. lawmakers it believes China's DeepSeek is attempting to replicate leading models via distillation, raising IP, security, and competitive concerns. The fight is shifting from benchmarks to model leakage controls, API abuse detection, and potentially new regulatory framing around 'model copying.'

A report suggests Meta is considering facial recognition features for smart glasses, pushing wearables toward real-time identification. That's a major capability upgradeand a direct collision with privacy regulation, consent norms, and platform governance.

xAI is facing legal pressure tied to alleged unpermitted generators and emissions concerns near a Mississippi facility, showing how AI scale-ups can collide with regulatory and community constraints. For AI operators, this is a reminder that power strategy is not just costit's compliance and reputation.

The EU has opened an investigation into X after reports that Grok generated sexualized imagery, escalating a product safety issue into a regulatory and platform governance risk. The incident highlights how generative AI features can become policy liabilities when safeguards fail under real-world use. For AI platforms, the takeaway is clear: content controls and enforcement now sit on the critical path to shipping.

Drugmakers are expanding AI use to accelerate clinical trial operations and streamline regulatory submissions, targeting time sinks like document drafting, data validation, and process coordination. The shift signals AI moving from experimentation to workflow infrastructure in heavily regulated environments. Success will depend on auditability, model governance, and compliance-grade traceability rather than raw model capability.

A U.S. House panel advanced a bill that would give Congress more authority over AI chip export controls, signaling rising political sensitivity around compute as a strategic asset. Export rules increasingly shape how AI infrastructure is built globally, affecting supply chains, vendor roadmaps, and regional availability. For businesses, this is a reminder that AI scaling is now constrained by policy as much as physics.

Eightfold AI is being sued in California over allegations that it helped companies compile reports to score job seekers without their knowledge. The case highlights how AI-driven hiring systems can become a legal and reputational liability when transparency and disclosure don't keep pace with automation. For HR tech buyers, the message is clear: AI scoring requires governance, explainability, and candidate-facing accountability.

China's regulator requires many AI systems to be filed into a public 'algorithm registry' before launchcreating an unusually transparent, government-curated index of the country's AI landscape. For builders, it turns safety and ideology requirements into a product gate, with structured disclosures across dozens of risk categories that effectively shape what gets deployed and how fast teams can iterate.

Growing concern over Grok's misuse in generating non-consensual intimate images has prompted the UK government to criminalise AI-generated deepfakes of this kind under new law. Regulator Ofcom is investigating Grok, and Elon Musk has publicly opposed what he calls authoritarian moves, intensifying the debate on platform accountability and AI safety. :contentReference[oaicite:20]{index=20}