EU AI Act general-purpose AI obligations become fully applicable on 2 August 2026 as the EU expands its AI Office with 38 new staff to enforce transparency and watermarking rules

The EU AI Act, described in the sources as the world's first comprehensive law regulating artificial intelligence, reaches a critical enforcement milestone on 2 August 2026, when the bulk of its provisions become fully applicable two years after the Act entered into force on 1 August 2024. From Sunday, AI companies must label or watermark AI-generated content under the Act's transparency obligations. Specifically, Article 12 of the EU AI Act becomes enforceable, covering transparency requirements for AI systems. To enforce these obligations, the European Union has expanded its AI Office in Brussels with 38 new staff, creating a specialist team to regulate AI companies globally, including addressing AI deepfakes, illicit imagery, and AI-facilitated hacking. Henna Virkkunen, described as the EU's tech sovereignty chief, emphasised the importance of building trust in AI as the new rules take effect. The timeline has been phased: prohibited AI practices and AI literacy obligations came into force in early 2025, rules governing general-purpose AI (GPAI) models (large-scale AI systems capable of performing a wide range of tasks, such as large language models) followed later that year, and Sunday's milestone marks the near-complete application of the framework. The Act applies to AI systems placed on the EU market or used within the EU, meaning it has direct compliance implications for UK-based technology companies, law firms deploying AI tools, and developers serving EU clients, regardless of where those entities are based. For UK practitioners, the EU Act's extraterritorial reach sits alongside a domestic landscape in which, as the sources note, a has not yet been passed, with sector regulators including the , , and the continuing to issue targeted AI guidance under existing powers.