EU AI Act Article 50 In Force: Compliance Framework Now Required for Every Client-Facing AI Tool

EU AI Act Article 50 transparency obligations took effect August 2. For mid-sized firms with EU clients, EU offices, or EU-accessible client-facing AI tools — AI research assistants, AI intake systems, AI client portals — disclosure architecture is now legally required. Providers must design AI systems to explicitly inform individuals when they are interacting with AI; machine-readable marks must be added to AI-generated content; deployers must disclose emotion recognition, biometric categorisation, and AI-generated text published on matters of public interest. The UK's approach is different: no AI legislation, sector-specific guidance, and voluntary compliance sandboxes (the AI Safety Testing Framework and AI Copyright Transparency Bill are in progress but not yet operative). The divergence creates a compliance asymmetry for UK-based mid-sized firms advising EU clients. For mid-sized firm IT directors and COOs: the practical compliance question is which client-facing systems involve AI-generated content, which involve emotion recognition or biometric categorisation, and whether those systems currently disclose AI involvement. Firms that reviewed their AI disclosure practices before August 2 are ahead; firms that have not need to map their client-facing AI systems against Article 50 obligations this week.