Society and philosophy

Google DeepMind is speeding up Gemini 4 while its new leader moves the lab away from AGI as its defining mission. Koray Kavukcuoglu said the model is in early post-training and that he hopes to release it “much earlier” than the end of the year. The timing matters: Google’s promised Gemini 3.5 Pro never arrived, while Anthropic and OpenAI have moved ahead with newer model families. The leadership change therefore looks like more than a change of pace. DeepMind, once organized around long-term research, is being run increasingly as Google’s product engine.

AI hacker houses are not just unusual living arrangements for ambitious founders. A New York Times investigation found multiple harassment and rape allegations connected to these Silicon Valley mansions, including 37 police reports involving AGI House since 2022. At least 17 were tied to the house’s large, expensive parties. The concern is larger than one address: these spaces bring together people who may later shape the AI industry, while operating with little visible accountability.

Jensen Huang has declared that artificial general intelligence has arrived. The Nvidia chief executive made the claim on Lex Fridman's podcast, in an episode framed around his company's $4 trillion valuation, and the exhibit is GPT-6 Astra. What makes the statement worth parsing is not the model but the measuring stick: Huang is not applying the research definition of AGI, and the definition…

OpenAI has released GPT-6 Astra, the first model the company has placed at the "critical" level of its preparedness framework — meaning that, given the right tools and access, it can find previously unknown vulnerabilities and build attack chains against well-defended systems without step-by-step human guidance. At the end of the press briefing announcing it, Greg Brockman said the AGI era had…

OpenAI presented GPT-6 Astra on September 3, 2026, and for the first time classified one of its own models as "critical" on cyber capability. Given the right tools and access, the company says, Astra can find unknown vulnerabilities and build chains of exploitation. It reports 100% on ExploitBench, 99.9% on ARC-AGI-3 and 97.6% on FrontierMath Tier 4 v2. API pricing starts at $10 per million…

OpenAI says its new model, GPT-6 Astra, is artificial general intelligence — by the company's own definition, "autonomous systems that outperform humans at most economically valuable work." The same model is the first OpenAI has ever placed in the "critical" category for cybersecurity capability, and the company has confirmed it shows a "substantial reduction in chain-of-thought…

OpenAI started handing GPT-6 Astra to enterprise customers on Thursday through Daybreak, a closed program, and framed it not as a better chatbot but as a model that drives a computer the way a person does. At a closed press briefing, Greg Brockman said the AGI era is arriving, and said he sees solid grounds for calling Astra itself AGI. OpenAI's own materials, shared in advance with…

Nscale has added Fidji Simo to its board of directors, a week after Bloomberg reported, citing unnamed sources, that the company is looking to raise up to $3.5 billion ahead of a planned IPO. Simo led AGI deployment at OpenAI, effectively the lab's second most senior executive, until she stepped down in July for health reasons; she continues to advise OpenAI part-time. Nscale, founded two…

GPT-6 Astra scored 62.7% on ARC-AGI-3, the benchmark that drops a model into unfamiliar game worlds without explaining the rules or the goal and makes it work them out by trial and error. GPT-5.6 Sol, its predecessor, scored 7.78%. Claude Opus 5 scored 30.16%. About six months ago, after ARC-AGI-3 shipped, François Chollet put the benchmark's saturation roughly a year out, depending on how…

Google DeepMind has launched an interdisciplinary institute to work on the big open questions about artificial general intelligence. The timing is the interesting part. DeepMind's stated position is that AGI is close — and the company is now standing up a research body to settle what it is that is close, because the field still has no commonly accepted definition of the term. DeepMind's own is…

Google DeepMind has launched an institute to widen the debate about AGI, and its first collection of essays goes straight at regulation. Demis Hassabis proposes a US-regulated standards body to assess the most advanced AI models, starting with developers voluntarily submitting them for review no later than 30 days before release. In a separate essay, DeepMind safety researchers Rohin Shah and…

There is a lot of argument around AGI right now: LLMs supposedly just predict the next word with some probability, so you cannot build "general AI" on them. The authors of The Missing Layer of AGI: From Pattern Alchemy to Coordination Physics from Stanford suggest looking at the problem differently. Their argument is that LLMs really do not hand you general intelligence out of the box — but…