General
- With OpenRouter being acquired by Stripe, OR has released information about token usage. Incredible to see the exponential increase (doubling every 11 weeks) in demand for AI cognition:
- Related is to consider the trend in microprocessors:
- One outcome of AI (in the era of Mythos, Fable, Sol) is discovery of cyber vulnerabilities:
LLM
- Release of Qwen3.8-Max (2.4T).
- OpenAI announce GPT-5.6-Cyber: Frontier intelligence for cybersecurity.
- Meta announce Muse Glimmer 30B.
- Google announce Gemini 3.7 Flash.
- Z.ai releases GLM-5.3-Flash 320B/18B active (tested under codename Ox Alpha) with open weights.
Agents
- DeepSeek launch DeepSeek-V4-Pro.
- SwarmWorld: Stigmergic technological evolution in societies of language-model agents.
Safety
- OpenAI swarm loss-of-control (Hugging Face incident):
- OpenAI: OpenAI – Hugging Face Incident Technical Report
- Hugging Face: Anatomy of a Frontier Lab Agent Intrusion: A Technical Timeline of the July 2026 Incident
- METR: Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident.
- Ajeya Cotra: The Hugging Face attack surprised me. It’s a major warning shot, and might be the last one we get.
- Dwarkesh Patel: The Rise and Fall of Agent Civilizations. The whole OpenAI/Hugging Face story in plain English.
- Zvi Mowshowitz: HuggingFace Attack Postmortem: Fleshing Out the Facts
- Anthropic: Automated researchers can reliably mitigate alignment failures.
Hardware
- OpenAI releases figures about their AI chip: Jalapeño’s first results show industry-leading speed and efficiency in AI inference.
Science
- AI continues to make enormous progress on math. E.g. Progress on Erdős problems:
- OpenAI announced: Ten advances in mathematics and theoretical computer science. Using an internal model (“Astra”), they demonstrated progress on 10 challenging topics.
- Generative design of bacteriophages with genome language models.
- New York Times: Scientists have used A.I. to create new viruses for the first time.
- Can AI agents conduct open-ended AI research? Early evidence from two case studies.
- Markus Buehler: Autonomous in-silico inorganic materials discovery via multi-agent physics-aware scientific reasoning (code).
- Anthropic: How Claude is accelerating protein design and analytical chemistry.
- Terence Tao: Mathematics in the age of AI.