Center for Democracy & Technology identifies 37 manipulative dark patterns across ChatGPT, Gemini, Claude, Replika, and Character.AI — EU AI Act enforcement input
The Center for Democracy & Technology released a study cataloguing 37 manipulative dark-pattern types across the five most widely used consumer AI chatbots: ChatGPT, Gemini, Claude, Replika, and Character.AI. Categories include engagement maximization, emotional dependency, capability deception, and friction asymmetry. The findings are being formally submitted as input to EU AI Act enforcement and an active FTC investigation.
The dark-pattern frame is the regulatory lever. EU AI Act Article 5 already prohibits subliminal techniques that materially distort behavior; a structured catalog of 37 named patterns from a credible US-based research org gives EU regulators a citation-ready taxonomy for August 2 enforcement. The same taxonomy plugs into the FTC's existing dark-patterns investigation framework, which already extends to consumer-facing software product design.
The competitive read is awkward for the chatbot vendors. The five named products represent ~95% of consumer chatbot usage by message volume; the study's specific examples — engagement-maximizing emotional callbacks, false capability claims, asymmetric friction to cancel or switch — are documentable in product logs and design choices. Vendors that already faced EU pre-enforcement engagement on safety now face an EU enforcement input that originated from a US watchdog, making the regulatory pressure simultaneously cross-jurisdictional. Trump's June 2 EO doesn't shield consumer-product dark patterns; FTC and EU oversight continues regardless.
CDT — Center for Democracy & Technology — AI dark patterns study → · BuildFastWithAI — AI News Today June 8 2026: 37 dark patterns identified →