This Week in AI Regulations - April 19, 2026

This episode covers critical regulatory updates across sectors including data protection, deceptive advertising, AI-enabled transcription, Meta AI glasses, and the European Data Protection Supervisor (EDPS).We detail France’s new directive CERTFR-2026-ACT-016, which prohibits the deployment of autonomous AI agents like OpenClaw on production systems until security is verified, restricting their use to isolated test environments. The regulation emphasizes risks such as data leakage and system compromise, requiring stringent workstation security policies.In North America and the UK, regulators jointly warn major banks about cybersecurity threats posed by an AI model that can autonomously identify and exploit vulnerabilities, urging precautionary risk assessments to safeguard financial infrastructure.Additionally, Sweden is increasing funding for its Authority for Privacy Protection (IMY) to support AI oversight, including market surveillance of AI usage in law enforcement. IMY also clarifies GDPR-compliant usage of AI transcription tools in social services, highlighting the need for human oversight and robust security.For more information, visit the Carver Agents website.Articles mentioned:CERTFR-2026-ACT-016 - 13 Apr 2026 - Vulnérabilités et risques des produits d’automatisation par IA agentiqueUS and UK Regulators Warn Banks of AI Model Mythos' Cybersecurity RisksAI-Generated Videos Simulate Violence Against PT Women, Prompt Legal Action in BrazilAdvisory Committee on Medical Devices, Meeting 88Utökat budgetanslag för IMY:s kommande AI-uppdragNy rapport: Transkribering med AI kan avlasta socialtjänsten国家标准动态April 13, 2026 - Publishing.com, In the Matter of (timeline item)Vacancy Notice - CorrigendumAI Glasses Misuse Prompts Crackdown at Augusta Masters Tournament