EU council reaches position on Chat Control (www.consilium.europa.eu)

The EU Council has agreed a negotiating position on a proposed regulation to prevent and combat child sexual abuse online. Key elements include mandatory risk assessments by online service providers, categorisation of services into low/medium/high risk, and powers for national authorities to require removal, blocking or delisting of child sexual abuse material (CSAM). The text creates a new EU Centre on Child Sexual Abuse to run a searchable reports database, maintain a library of CSAM indicators for industry use, assist victims, and share data with Europol and national law enforcement. Providers in the high‑risk category can be ordered to contribute to development of mitigation technologies; non‑compliance can trigger penalties. The Council also proposes to make permanent the current temporary exemption that allows voluntary scanning of communications for CSAM. For the AI/ML community this raises immediate technical and policy implications: expect stronger regulatory pressure to deploy automated detection tools (hashing, perceptual fingerprinting, ML classifiers, multimodal detectors), integrate reporting pipelines, and contribute to shared indicator databases. Model builders and platforms will need robust CSAM filtering and dataset governance to avoid training on illegal content and to prevent generative models from producing illicit imagery. The proposal also intensifies the trade‑off between effective detection and encryption/privacy — driving interest in on‑device detection, metadata risk‑mitigation, privacy‑preserving ML and interoperable indicator formats. The Council text now goes to trilogue with the European Parliament; details and limits on scanning mandates will determine technical obligations.