Hallo zusammen, sofern nicht selber schon gesehen/-lesen hier [0] FYI "New TokenBreak Attack Bypasses AI Moderation with Single-Character Text Changes" --> siehe dayu auch [1] --> "... To defend against TokenBreak, the researchers suggest using Unigram tokenizers when possible ..." VG Bernd [0] https://thehackernews.com/2025/06/new-tokenbreak-attack-bypasses-ai.html [1] https://arxiv.org/abs/2506.07948 -- Bernd Kohler IT Center Abteilung: Netze RWTH Aachen University Wendlingweg 10 52074 Aachen Tel: +49 241 80-29793 Fax: +49 241 80-22666 kohler@itc.rwth-aachen.de www.itc.rwth-aachen.de Social Media Kanäle des IT Centers: https://blog.rwth-aachen.de/itc/ https://www.facebook.com/itcenterrwth https://www.linkedin.com/company/itcenterrwth https://www.youtube.com/channel/UCKKDJJukeRwO0LP-ac8x8rQ