The EU AI Act entered general application on August 2, 2026, although some high-risk deadlines have been pushed to 2027 and 2028 under the new EU agreement known as the Digital Omnibus.
August 2, 2026 was the general application date of the EU AI Act. From that date the Article 50 transparency requirements (such as labeling AI-generated content), the European Commission's fining powers over general-purpose AI (GPAI) models, and the full powers of national market-surveillance authorities became officially active.
However, on May 7, 2026 EU institutions reached a provisional agreement called the "Digital Omnibus on AI", which rewrote part of the implementation timeline. Under it, the deadline for standalone high-risk systems (Annex III — covering uses such as hiring decisions, credit scoring and biometric identification) was postponed sixteen months to December 2, 2027.
The deadline for high-risk systems embedded in products (Annex I) was also moved to August 2, 2028. Meanwhile, the transition period for Article 50(2) requirements for generative systems placed on the market before August 2, 2026 runs until December 2 of that year.
On penalties, violations involving prohibited AI practices can be fined up to 35 million euros or seven percent of global annual turnover; other operator violations carry fines up to 15 million euros or three percent of global turnover.
The practical point for businesses outside the EU that offer products or services to European users: the transparency requirements are already active and must not be ignored, but there is more time (to 2027 and 2028) for full compliance of high-risk systems — time that should be used for planning and risk assessment, not for pointless delay.
This note is a summary and rewrite based on the report below; see the original source for the full report.
Latham & WatkinsThe 2026 report by the Cambridge Centre for Alternative Finance (CCAF), with BIS, IMF, WEF and others, shows 81% of financial firms use AI and software engineering is the most mature use — yet concern over lost human oversight and weak explainability is serious.
According to a SiliconANGLE report on September 18, 2026, a week that saw multi-billion-dollar AI infrastructure investments alongside safety tools to rein in rogue agents — the AI-agent industry's paradox.
Research by Medrano, Verma and Chhabra shows that in a production-like RAG system, RAG Fusion raises recall, but after re-ranking and context truncation much of the gain disappears and precision even dips slightly.
