Commentary · Oct 16, 2026
EU Artificial Intelligence Act Enforcement Begins With First Prohibition Notices Issued to Social Media Recommendation Systems
The European AI Office issued its first formal prohibition notices under the EU Artificial Intelligence Act on 25 August 2026, targeting social media content recommendation systems operated by two major platforms that the Office found to employ 'subliminal techniques' capable of influencing user behaviour without conscious awareness. The notices require modification or shutdown of the affected systems within 90 days and open the path to fines of up to 35 million euros or 7 percent of global annual turnover.
Contents
Key takeaways
- First prohibition notices issued to social media recommendation systems, not generative AI.
- 90-day compliance window before potential fines of EUR 35 million or 7% of global turnover.
- European AI Office found systems used 'subliminal techniques' meeting the prohibition threshold.
- Both companies have indicated they will challenge the notices through EU administrative courts.
- AI Office published guidance on the evidence standard for subliminal technique findings.
Legal Basis
Article 5(1)(a) of the AI Act prohibits AI systems that 'deploy subliminal techniques beyond a person's consciousness or purposely manipulative or deceptive techniques, with the objective, or the effect, of materially distorting the behaviour of a person or a group of persons in a manner that causes or is reasonably likely to cause significant harm'. The AI Office's prohibition notices found that the flagged recommendation systems met this definition based on internal testing and a review of the companies' own design documentation disclosed during a supervisory investigation.
The companies, which the AI Office declined to name pending the administrative review process, are understood to be a major US-headquartered video sharing platform and a European social network. Both operate recommendation systems that adjust content ordering based on granular behavioural signals including micro-dwell time and scroll velocity.
Industry Reaction
The prohibition notices have triggered significant concern among technology trade associations, which argued in public statements that the AI Office's evidence standard is too broad and could capture any personalisation algorithm. The Computer & Communications Industry Association called for clarification of the 'subliminal' threshold and urged the AI Office to publish detailed technical criteria before expanding enforcement.
Broader Implications
The AI Office has indicated it is reviewing recommendation systems operated by additional platforms and is conducting a sector-wide inquiry under Article 68 of the Act, which allows it to require all providers in a market segment to submit technical documentation for review. Analysts expect further prohibition notices to follow within six months.
Sources
Related publications
The region’s AI laws stop where the drones start
Kazakhstan and its neighbours are regulating civilian AI while buying drones. The rules for autonomous weapons are being written in Geneva, where Armenia and Azerbaijan are not parties.
Switzerland and EU Finalise Bilateral Framework on Digital Markets and Data Governance
Switzerland and the European Union initialled a bilateral Digital Markets and Data Governance Agreement on October 19, linking Switzerland to the EU's Digital Markets Act, Digital Services Act, and European Data Governance Act for the first time. The deal, part of the broader Bilateral III framework concluded in June 2026, gives Swiss digital service providers continued access to EU markets in exchange for Switzerland dynamically adopting EU digital regulation. Swiss data protection authorities will sit as observers in key EU enforcement committees.
China's Xinjiang Cotton Certification Scheme Disrupts EU Textile Trade Under Forced Labour Regulation
The EU's Forced Labour Regulation enforcement unit banned 47 textile product lines from Chinese suppliers on October 16 following evidence that their cotton inputs were traceable to Xinjiang production zones covered by the regulation's rebuttable presumption of forced labour. China's government-backed Xinjiang Cotton Certification Scheme, launched in June 2026 to supply traceability documentation to EU importers, has been assessed as failing the EU's evidentiary standards. Major European fashion retailers face supply chain disruption heading into the peak winter season.