EU set to water down landmark AI act after Big Tech pressure (www.ft.com)

Reports indicate the European Union is preparing to water down its landmark AI Act following intense lobbying from major technology firms. Negotiators are said to be softening key provisions — for example narrowing the definition of “high‑risk” systems, carving out or delaying obligations for large foundation and general‑purpose models, and reducing transparency, auditing and data‑governance requirements. The changes are framed as compromises to preserve innovation and competitiveness, but are reportedly driven by industry pressure to limit compliance burdens and liability for cloud and platform providers. For the AI/ML community, the shift matters because it changes the regulatory baseline that developers, researchers and deployers must meet across the bloc. A weaker law could leave gaps in oversight of large-scale models (safety testing, red‑teaming, provenance and dataset documentation), slow adoption of mandatory risk‑mitigation practices, and favor incumbents able to influence policy and absorb reputational risk. Conversely, fewer prescriptive rules might reduce short‑term compliance costs for startups and labs. Technically, the outcome will determine whether requirements for model transparency, explainability, logging, external audits and pre-deployment risk assessments become binding — shaping how models are built, documented and deployed in Europe for years to come.