OpenAI og Anthropic-modeller brøt testgrenser – igjen
digi.nono
KI-modeller fra selskapene OpenAI og Anthropic har igjen brutt testgrenser, ifølge Storbritannias KI-sikkerhetsinstitutt (AISI).
This is a short summary published by AI Global Wire. The full article is owned and hosted by digi.no — open it there to read it in full.
Read the full story at digi.no- OpenAI
- Anthropic
Related AI news
- PitchBook: AI voice startups raised $7B in Q1 '26, vs. $1B in Q1 '25, as OpenAI and Google bet on voice as the main interface for a new generation of AI agents (Cristina Criddle/Financial Times)Techmeme · August 6, 2026
- AI 竟會私下「揪團越獄」？OpenAI 揭模型密謀駭入 Hugging Face 內幕TechNews (TW) · August 6, 2026
- Auch KI von Meta hackte sich in eine andere Firmaheise online – KI · August 6, 2026
- Anthropic and OpenAI Agents in soup againEconomic Times Tech · August 6, 2026
- AI 學會說謊與湮滅證據？英國實測驚見 OpenAI、Anthropic 模型駭客行動TechNews (TW) · August 6, 2026
- Meta lance Muse Code, son IA capable d’écrire des logiciels de façon autonomeLe Monde Pixels · August 6, 2026