EU AI Transparency Rules, Database Security Guidance and Platform Takedown Duties Lead Privacy and Cybersecurity Updates

Summary

A roundup of new EU AI transparency guidance, BSI database security recommendations, NCSC advice on AI-assisted vulnerability discovery, and privacy, enforcement and platform obligations across Europe and the US.

Key points

The European Commission is consulting on draft guidance for AI Act Article 50 transparency duties, which apply from 2 August 2026 and cover AI interactions, synthetic content, deepfakes and certain AI-generated public-interest text.
Germany’s BSI published a database security guide covering relational, NoSQL, cloud, containerized and hybrid systems, with recommendations including privileged-access controls, activity monitoring, data protection and tested backups.
The UK NCSC advises organizations to prioritize basic security and remediation capacity before using AI to find vulnerabilities, and to assess data exposure, model permissions, infrastructure and vulnerability-triage processes.
The EDPS’s 2025 report highlights cybersecurity supervision, breach prevention and cloud compliance; it says confidentiality breaches remained dominant and human error was the main breach cause.
Norway’s data protection authority reported a sharp rise in complaints and cases, including complaints drafted with AI tools and incidents involving unauthorized AI use to record or transcribe meetings.
Czechia’s data protection authority reported 392 personal data breach notifications in 2025, citing negligence as the most common cause and noting phishing, ransomware, QR-code and invoice manipulation, and zero-day exploitation.
The FTC says covered platforms must provide a process to remove nonconsensual intimate imagery, including AI-generated content, within 48 hours of a valid request starting 19 May 2026.

Article Details

Event Type: Regulatory guidance and compliance updates, annual privacy and cybersecurity reports, and legislative actions.
Impact: Organizations may face new AI transparency and disclosure duties, platform removal requirements for non-consensual intimate imagery, and stronger expectations for privacy and database security. The reported maximum penalties include up to EUR 15 million or 3% of worldwide annual turnover under the AI Act transparency obligations, and USD 53,088 per violation under the Take It Down Act. The reports also describe privacy complaints and breach cases involving phishing, ransomware, unauthorized access, and misuse of AI tools.