https://meridianradar.ai/Daily plain-English AI regulatory radar.https://meridianradar.ai/brief/2026-08-08/eu-transparency-duties-are-live-while-high-risk-rules-wait-the-two-track-reality2026-08-08The Digital Omnibus deferred Annex III high-risk obligations to December 2027, but Article 50 transparency duties, GPAI enforcement, and the full penalty regime became applicable August 2, 2026 — with a marking grace period ending December 2, 2026.https://meridianradar.ai/brief/2026-01-01/texas-traiga-takes-effect-nist-compliance-becomes-a-legal-shield2026-01-01The broadest US state AI law in force pairs enumerated prohibited harms and intent-based liability with a safe harbor for substantial NIST AI RMF compliance.https://meridianradar.ai/brief/2026-05-28/colorado-repeals-its-ai-act-before-effect-narrower-replacement-lands-january-2022026-05-28The first comprehensive state algorithmic-discrimination law was repealed pre-effect; its replacement requires pre-use notices, 30-day explanations, and human review, with AG-only enforcement from January 1, 2027.https://meridianradar.ai/brief/2026-08-02/ai-act-penalty-architecture-fully-in-force2026-08-02The Act's full fine regime became applicable: up to 3% of global turnover for most violations, with the Article 50 transparency regime enforceable and a marking grace period to December 2, 2026.https://meridianradar.ai/brief/2026-07-15/china-s-companion-ai-measures-take-effect2026-07-15Five agencies' interim measures for anthropomorphic AI interaction services became effective — the first dedicated national regulation of human-relationship-simulating AI anywhere.https://meridianradar.ai/brief/2026-06-24/ofcom-opens-investigations-into-ai-companion-services2026-06-24Under the Online Safety Act, Ofcom confirmed open investigations into AI companion products — the Western counterpart to the category China regulated weeks later.https://meridianradar.ai/brief/2026-04-30/federal-banking-agencies-retire-sr-11-72026-04-30Updated model-risk guidance replaced the fifteen-year-old letter, explicitly excluding generative and agentic AI from scope — with separate expectations promised and an interim governance gap named.https://meridianradar.ai/brief/2026-03-18/ftc-lands-18m-voice-ai-settlement2026-03-18Operation AI Comply's largest monetary outcome: an $18M judgment and a permanent business-opportunity marketing ban over deceptive AI earnings claims — Section 5 reaching AI conduct without new rulemaking.https://meridianradar.ai/brief/2026-01-22/korea-s-ai-framework-act-enters-into-force2026-01-22Asia-Pacific's first comprehensive mandatory AI law: risk-based duties, high-impact obligations, generative-output labeling, and a domestic-representative requirement with extraterritorial reach.https://meridianradar.ai/brief/2026-01-06/fda-narrows-premarket-review-for-clinical-decision-support2026-01-06Revised guidance let single-recommendation CDS proceed under enforcement discretion where clinicians can independently review the logic — transferring the evidentiary burden from agency to organization.https://meridianradar.ai/brief/2025-12-10/executive-order-signals-national-ai-framework-and-preemption-posture2025-12-10A December order signaled intent to establish a single federal framework and pressure the state patchwork, including through litigation — leaving state law binding while the fight proceeds.https://meridianradar.ai/brief/2025-09-01/china-s-content-labeling-mandate-becomes-operative2025-09-01Visible and embedded marking of AI-generated material became mandatory under a binding technical standard — the world's most concrete synthetic-content transparency implementation, previewing the EU's Article 50 and US state rules.