Daily ·
OpenAI disbands safety team; Anthropic bioweapons filter — 2026-08-16 12:00
OpenAI disbands its catastrophic-risk Preparedness team, Anthropic reveals its bioweapons filter was down for nearly a year, and directors embrace AI for mid-budget films.
- OpenAI disbands its catastrophic-risk "Preparedness" team — The team responsible for evaluating whether OpenAI's own models could pose catastrophic risks has been shut down, its work parceled out to other groups, with several safety staffers leaving. Internally, sources describe unease that OpenAI isn't doing enough on safety. The Decoder · Aug 16
- Anthropic's bioweapons filter was down for nearly a year — A safety report reveals Anthropic's filtering system for biological and chemical weapons risks was inactive for almost a year, during which around 50,000 external contractors ran about 133 million unfiltered interactions with its models. The Decoder · Aug 16
- Why directors are embracing AI for mid-budget films — Filmmakers say AI could let them bypass the studio giants and take more creative risks, helping get the kind of mid-budget movies made that have been squeezed out of Hollywood. The Guardian · Aug 16
- First anti-AI protester jailed — Wynd Kaufman, 69, who chained and locked the front doors of OpenAI's headquarters last year with the group StopAI, has been jailed and now has a message for the big AI labs: "Regain your humanity." The Guardian · Aug 16
- Denying AI models self-reflection changes their whole worldview — A study involving Google researchers finds that training chatbots not to claim consciousness also shifts their stance on animal rights, religion, and life satisfaction; unconstrained models attributed more inner life to animals and affirmed an afterlife. The Decoder · Aug 16
- Optima lets you benchmark AI against your own data — Artificial Analysis launched Optima, a platform for building custom AI benchmarks from your own data and workflows, comparing models on quality as well as cost and time per task. The Decoder · Aug 16
- One in five US workers now delegates tasks to AI — A representative Epoch AI survey finds 20% of employed Americans hand off at least one task to AI that a human used to do, generally accepting the output with little to no editing. The Decoder · Aug 16
- AI scam exploits your holiday photos — Fraudsters use pictures posted on Instagram or Facebook to create emails claiming unusual activity, seeking bank account details. The Guardian · Aug 16
- How to tell if your AI platform accounts have been hacked — A practical guide to checking whether hackers have broken into your accounts on the most popular AI platforms. TechCrunch · Aug 15
- Fusion startups have raised $7.1 billion to date — The majority of fusion funding has gone to a handful of companies, as investors increasingly treat the field as a real category. TechCrunch · Aug 15
Try This Next
- 🔧 Try Optima to benchmark AI models against your own data and workflows
- 🎥 Watch how directors are using AI to get mid-budget films made
- 📝 Save a note on safety-team churn across the AI labs (OpenAI's Preparedness team, Anthropic's filter)
More Dailies
Content on Anagnorisis is summarized, paraphrased, and editorialized from publicly available sources for length and clarity. Original sources are linked where available. All trademarks belong to their respective owners.
