EU AI Act coverage highlights enforcement and platform pressure

EU AI Act-related coverage spans 8 AI Weekly issues and 63 live stories from the last 30 days, with 17 total mentions and activity tracked since 2026-04-18. Recent items focus on Brussels’ enforcement agenda, including Meta’s automated ad review rejecting a Barcelona “Room of One’s Own” play as “social issues” content, Spain’s AEPD logging a data-breach notification handled end-to-end by an autonomous AI agent, and EU action placing ChatGPT under the DSA as the first AI service while adding Reddit and Roblox to the VLOP list.

Transparency and labeling remain central. Anthropic signed the EU AI Act Article 50(2) code and added C2PA marks and watermarks to Claude output, while Article 50 labeling and watermark rules were listed as taking effect August 2, with fines reaching €15M or 3% turnover. The European Commission also published transparency guidelines for chatbots and deepfakes ahead of the August 2 enforcement deadline.

Broader enforcement and sovereignty themes include plans affecting Google search data, Android access for rival AI assistants, cloud gatekeeper scrutiny of AWS and Azure, and European efforts to build sovereign AI capacity. The live list also highlights tensions over access to Anthropic’s Mythos models, EU cybersecurity planning, and investment in open and regional AI infrastructure.