European Parliament Extends Voluntary CSAM Detection Rules Until August 2027 Amid Permanent Framework Negotiations

The European Parliament voted on November 3, 2026 with 458 in favour, 103 against and 63 abstentions to extend a derogation from the ePrivacy Directive allowing voluntary detection of child sexual abuse material online until August 3, 2027. The current exemption was set to expire on April 3, 2026, and the extension provides a bridge until agreement is reached on a permanent legal framework to prevent and combat child sexual abuse online. Under the extension, technology used for voluntary detection should only apply to material already identified as CSAM or flagged as potential CSAM by a user, trusted flagger or organisation, with MEPs emphasising that voluntary measures must be proportional and limited to what is necessary. Parliament has been ready for negotiations on the permanent framework since November 2023, and talks have been ongoing since the Council adopted its position in November 2025.

Voluntary detection frameworks depend on platforms choosing to act and on the technical and legal capacity to do so, leaving gaps where providers decline participation, operate under end-to-end encryption, or face uncertain liability. Guardii provides detection at the edge of the conversation itself, deployed by the organisations responsible for the child, including schools, sports bodies, and safeguarding agencies, rather than relying on platform cooperation. The temporary derogation permits identification of previously hashed material and user-flagged content; Guardii extends that operational range by detecting grooming, sextortion and CSAM distribution in real time through behavioural and linguistic pattern recognition, identifying threat escalation, coercion and acute distress even in novel or synthetic content that has not yet been hashed or reported. As the EU navigates the privacy and enforcement trade-offs inherent in platform-wide scanning, targeted detection offers a model that protects children without requiring mass surveillance, instead surfacing the specific conversations where harm is occurring to the adults authorised to respond.