The article claims: > your private communications can be searched by error-prone artificial intelligence technologies. Although these algorithms are meant to s

The article claims: > your private communications can be searched by error-prone artificial intelligence technologies. Although these algorithms are meant to search for potential child pornography and grooming, up to 86% of the correspondence reported to the police is not criminally relevant and users are falsely being reported – including many minors. > This takes place in a fully automated process and using error-prone “artificial intelligence”. If an algorithm considers a message suspicious, its content and meta-data are disclosed automatically and without human verification to a private US-based organization and from there to national police authorities worldwide. The proposed legislation contains: > (ac) the provider of the number-independent interpersonal communications services ensures human oversight of, and, where necessary, intervention in the processing of personal data using technologies falling under this Regulation, and ensures that no report of material not previously identified as child sexual abuse or of solicitation is sent to law enforcement authorities or organisations acting in the public interest against child sexual abuse without prior human confirmation > (b) the technology used is in itself sufficiently reliable in that it limits to the maximum extent possible the rate of errors regarding the detection of content representing child sexual abuse, and where such occasional errors occur, their consequences are rectified without delay; My conclusion as a layman: This is problematic, but if you wanted to write "good" anti-child-pornography legislation, this might be close to it?