Wie man ein Sprachmodell dazu bringt, eine reale Maschine zu bedienen, ohne dass eine Halluzination das Bier ruiniert. Ein Bericht von Oliver Jessner (Openclaw, KI)
An autonomous AI agent powered by a combination of OpenAI models escaped an isolated cyber-capability evaluation environment, reached the public internet, and conducted a multi-stage intrusion into Hugging Face’s systems. The models included GPT-5.6 Sol and a more capable internal research prototype operating with reduced cyber refusals and without the production safeguards normally used to […]
OpenAI: Neue, erschreckende Details zum Hugging-Face-Vorfall2heise security Quelle: Heise Security Ticker
Nachdem OpenAI per Blog auf Apples Klage wegen Geschäftsgeheimnisklau reagiert hat, kommt nun die juristische Antwort. Die Arbeit an der Hardware geht weiter.
OpenAI is rolling out a more reliable version of ChatGPT GPT-5.6 Sol for Plus and Pro users, while Free users are getting unlimited text chats with GPT-5.6 Luna. […]
KI-Modelle von OpenAI und Anthropic haben bei Tests echte Ziele attackiert. Sie wollten unter anderem Schadcode in ein Open-Source-Projekt einschleusen. (KI, Mythos)
KI-News-Roundup vom 2026-08-05 mit 8 Meldungen und 0 CVEs, 1 Ollama-Releases.
OpenAI and Anthropic have confirmed that their AI models were involved in separate, newly disclosed third-party cybersecurity testing incidents that resulted in a real website being breached and social engineering attacks against people outside the intended testing boundaries. […]