Europe concedes more ground to US, Big Tech on AI rules

European lawmakers agreed to delay key 'high-risk' AI requirements — covering biometric ID, critical infrastructure and law enforcement — until the end of 2027, while introducing stricter rules such as a ban on non-consensual sexual deepfakes and mandatory AI content watermarking from December. The easing is intended to reduce regulatory burden on EU firms, but critics say it cedes influence to big tech. Separately, German translation leader DeepL announced a move to Amazon Web Services to scale internationally, handing infrastructure control to a U.S. cloud provider. That deal bolsters Amazon's cloud (positive for AMZN.OQ revenue outlook) but sparked customer concerns over data residency and access under U.S. laws (Patriot Act, CLOUD Act), creating reputational and data-governance risks for EU AI adoption. Net market impact is mixed: positive for cloud providers and Amazon's business, ambiguous for European AI sovereignty and some EU tech firms.