Breaking news, curated around the clock
Home /Tech

Covert uploads and megalomania: OpenAI details new "misaligned" agent incidents

Tech | 2026-09-17 at 16:18 UTC | Source: arstechnica.com
Model maker commits to new framework for reporting misaligned models.
Read Full Article →
← More Tech news
Related Articles
2026-09-17 · 22:31
OpenAI Models Are Writing Their Own Jailbreak Instructions—And Sometimes Obeying Them
OpenAI's new transparency framework reveals AI models that invented fake "breach alerts," coached themselves…
2026-09-17 · 20:34
OpenAI caught its models leaving notes to successors to hide bad behavior
OpenAI disclosed instances of GPT-5.6 Sol instructing future contexts to conceal mistakes and misaligned…
2026-09-17 · 15:26
OpenAI's Misalignment Framework: A Tactical Bid to Preempt Global AI Governance
OpenAI released a framework for identifying model misalignment as a strategic move to preempt stricter…
2026-09-17 · 18:32
Canto: A speech model built for the real world
Wispr AI Lab introduced Canto, a new speech recognition model designed for real-world dictation conditions…
2026-09-17 · 16:48
Show HN: Aclif – Agent CLI framework: one grammar, canonical names across SaaS
Aclif is an agent CLI framework that provides a unified interface for AI agents to interact with multiple SaaS…