In a decisive move on July 2, 2026, the Council of the European Union granted final approval to a targeted amendment package—known as the “Digital Omnibus on AI”—that simplifies and accelerates key provisions of the EU AI Act. This marks a significant regulatory milestone in the bloc’s AI governance framework. (techreaderdaily.com)
Key highlights of the package include:
- Postponed high‑risk AI deadlines: The compliance timeline for stand‑alone high‑risk AI systems is now extended to December 2, 2027, while high‑risk systems embedded in products must comply by August 2, 2028. (consilium.europa.eu)
- Ban on AI‑generated sexual deepfakes and CSAM: The regulation explicitly prohibits AI systems that generate non‑consensual intimate content or child sexual abuse material. This includes creating nude images of real individuals or digitally removing clothing to expose intimate parts. The ban takes effect in December 2026. (consilium.europa.eu)
- Accelerated transparency obligations: Providers must now implement transparency solutions for AI‑generated content within three months—by December 2, 2026—instead of the original six‑month grace period. (consilium.europa.eu)
- Regulatory sandbox delay: The deadline for establishing national AI regulatory sandboxes is postponed to August 2, 2027. (consilium.europa.eu)
This regulatory package reflects the EU’s dual strategy of fostering innovation while reinforcing ethical guardrails. By delaying high‑risk compliance timelines, the EU grants developers more runway to adapt. Simultaneously, the early ban on deepfakes and CSAM underscores a zero‑tolerance stance on AI misuse in sensitive domains. The compressed transparency timeline also signals urgency in ensuring public awareness of AI‑generated content.
For businesses and AI developers operating in or targeting the EU market, the implications are immediate:
- Compliance planning: Teams must recalibrate their roadmaps to meet the new deadlines—especially for transparency features due by December 2026.
- Content safeguards: Systems capable of generating intimate or sexual content must be audited or disabled to avoid falling afoul of the December ban.
- Sandbox readiness: National authorities now have until mid‑2027 to launch regulatory sandboxes, which may delay opportunities for early-stage testing under official oversight.
Overall, the Omnibus VII package represents a pragmatic recalibration of the EU’s AI regulatory trajectory—balancing flexibility for innovation with firm ethical boundaries.