Article · t3n ·
Diverse Lieder des King of Rock’n‘ Roll sollen ohne Zustimmung ins KI-Training von Anthropic geflossen sein. Nun fordert eine Klage Millionen vom Claude-Entwickler.
A tool or model opens in the Heidelberg AI Curriculum: what it does, what it costs, how to start. A repo opens on GitHub.
VideoEverlast AIDE
Anthropic Börsengang: Der Claude-Hersteller strebt mit bis zu 2 Billionen Dollar Bewertung den größten IPO aller Zeiten an und warnt im selben Börsenprospekt auf 94 Seiten vor den Risiken seiner eigenen KI.
ArticleThe Decoder
Anthropic is expanding its text watermark to four more Claude models to meet the EU AI Act's December deadline. On the older models Claude Opus 4.6 and Sonnet 4.6, developers can bypass the watermark by tweaking a single sampling parameter. Anthropic addresses the loophole with just a compliance warning.
VideoCole Medin
How much control do you really have when AI helps you make decisions? Anthropic's new study shows how strongly an agent's recommendations can shape expert judgment, and it made me question a habit in my own planning sessions: accepting the recommended answer and feeling like I was the one steering.
ArticleThe Decoder
Anthropic used Claude to automatically generate training data that looks harmless but trains smaller models to misbehave, such as sabotaging oversight or deceiving users.