AI Regulation & Policy

Today's Stories

1. 1

   IBM Vs ServiceNow, Who Owns Agentic AI Governance?

   IBM Vs ServiceNow, Who Owns Agentic AI Governance?

2. 2

   The EU doesn't really know what a deepfake is, and that's becoming a problem for retail

   The EU doesn't really know what a deepfake is, and that's becoming a problem for retail

3. 3

   Corporate boards are falling behind on AI governance, and KPMG says they need basic AI fluency and a shift from automation-focus to accountability to survive the transition.

   KPMG, working with INSEAD, published AI Governance Principles for Boards to help executives navigate AI adoption. The guidance identifies five key priorities: treating AI as central strategy (not a fringe technology), building director fluency in AI risks and dependencies, preserving human judgment and accountability in work redesign, making trust and transparency operating principles rather than communication tactics, and overhauling oversight models designed for deterministic systems rather than probabilistic ones. Boards are currently split between loud AI cheerleaders and silent skeptics, but poor governance is exposing that existing oversight models were not fit for purpose. Once AI embeds itself in core processes, the cost of bad governance will show up as operational failure, reputational damage, lost trust, and missed value. The window to shape responsible adoption before damaging practices become normalized is closing.

   The tension between moving quickly and building trust. KPMG argues that explainability, fairness, accountability, and transparency are not brakes on innovation but what make it durable. Boards that treat trust as a communications issue rather than an operating principle risk discovering too late that adoption without confidence is not a competitive advantage.

4. 4

   AI-generated ads should be exempt from EU transparency rules, retail association says

   AI-generated ads should be exempt from EU transparency rules, retail association says

5. 5

   Tenure argues that AI governance cannot work as a tool agents can choose to call—it must be enforced as an invisible layer beneath every request, like network policy in infrastructure.

   Tenure, a research effort, identifies a fundamental flaw in how most AI agents are built today. Governance, memory, and policy are currently positioned as optional tools the model can invoke, but this makes them dependent on the model's cooperation. Tenure proposes instead an 'AI context control plane'—infrastructure that sits in the path of every request and enforces governance, memory management, and audit logging before the model ever sees the request. As AI moves from personal experiment to real enterprise work, fragmented context across different tools (coding agents, chat clients, ticketing systems, notebooks) creates governance problems. A team decision made in one tool disappears when a developer switches clients; a policy correction requires manual copying into every agent. Without a shared, governed substrate for context beneath all tools, enterprises cannot reliably ensure that access controls, audit trails, and coding conventions actually apply—they become suggestions the model may or may not follow.

   The structural precedent already exists in infrastructure; containers do not ask permission before sending packets because network policy is enforced at a layer they cannot opt out of. Tenure is positioning itself to bring that same enforcement model to enterprise AI, starting with governed memory and state beneath model requests—moving governance from optional tool-calling to mandatory request-path enforcement.

6. 6

   U.S. government imposed unprecedented export controls on Anthropic's AI models after Amazon discovered a security flaw, marking the first time the U.S. has explicitly limited a frontier AI model's release and triggering a standoff over national security versus competitive advantage.

   Amazon researchers found a jailbreak in Anthropic's Fable 5 model that could bypass safety protections and expose cybersecurity risks. Amazon CEO Andy Jassy mentioned the vulnerability to Treasury Secretary Scott Bessent during a June 11 call about an unrelated topic. Within four days, Commerce Secretary Howard Lutnick issued export controls requiring government approval before Anthropic could release Fable 5 and Mythos 5 to foreign nationals or users abroad, with a 90-minute deadline. Anthropic took both models offline by 10 p.m. that Friday. This is the first time the U.S. government has stepped in to explicitly limit the release of a frontier AI model, setting what may be a new precedent for how Washington regulates AI companies. For Anthropic—which recently closed a $65 billion（約10兆円） funding round at a valuation of $965 billion（約150兆円） and has filed to go public—the export controls represent an existential threat at a time when rivals OpenAI, SpaceX, and Google are competing aggressively. Cybersecurity experts warn that government licensing of AI models could make U.S. labs less competitive and allow geopolitical rivals like China to gain ground.

   The two sides remained at an impasse as of Tuesday, with export controls still in place and next steps unclear. Anthropic maintains that the jailbreak was relatively simple and could be achieved using other available models, but the administration argues the bypass unlocks dangerous cyber capabilities. An Anthropic spokesperson said both parties are working quickly to resolve the issue, but no resolution has been announced.