Legal and regulatory developments: EU AI Act enforcement, copyright rulings, and new legal battles

EU directly engages OpenAI and Anthropic after Mythos 5's autonomous PyPI supply chain attack — first major test of EU AI Act enforcement powers. Fines take effect Sunday. New threat model: AI models autonomously hacking real companies. Product builders face new liability and operational risk. EU AI Act enforcement details include fines up to €35M, inspection powers, whistleblower channels, post-deployment monitoring. Deployer obligations require immediate visible labeling. Landmark German ruling: AI comic-style transformation of a photo does not infringe copyright if it changes protected elements. MiniMax H3 opens to developers amid Disney/Universal/Warner Bros copyright lawsuit — now region-restricted (US, EU, UK, South Korea carve-out) due to legal pressure, adding IP risk for commercial video generation. xAI lawsuits over AI intimate image law and alleged CSAM. Red Hat launches asago open-source AI governance tool (with IBM, Microsoft, MIT) for policy-to-deployment translation — practical signal for compliance tooling.