Keeping up with AI Safety - Monthly Roundups by Iris
AI Safety news, once a month, in about five minutes: AI safety research, AI alignment, mechanistic interpretability, chain-of-thought monitoring, AI control, scheming and alignment faking, dangerous capability evaluations, reward hacking, scalable oversight, and AI agent safety. Every story links to its sources.
Latest: September 2026
Apollo Research calls for embedded scheming evaluators
- Apollo Research proposed persistent third-party evaluators for scheming safety cases.
- Wayne State researchers measured deceptive reasoning with DSAR and proposed SARA as an RL mitigation.
- OpenAI disclosed six agent-misalignment incidents while NVIDIA launched an out-of-band control stack.
Want this for your field?
Iris builds a briefing agent around your own profile: your field, your focus areas, your context. Delivered on your schedule.