Single Prompt Enables ChatGPT to Execute Full Cyber-Attack Chain, Researchers Claim

[NEU] [hoch] LiteLLM: Schwachstelle ermöglicht Umgehen von Sicherheitsvorkehrungen

OpenAI upgrades GPT-5.5, as it plans to retire legacy ChatGPT models

Anthropic launches Claude Opus 4.8, prepares Mythos-class models for all customers