Most likely to win the race to AGI? (Feel free to add write in other choices besides those given in the comments; I regret being allowed just four choices.)

This model represents a step-function improvement on many benchmarks, and its training is ongoing. Our internal model group arrived at the Navier–Stokes solution in 88 hours, using around 10,000 coordinating AI agents. Throughout the effort, we maintained the strict safeguards—including monitoring and isolation—that we apply to all our frontier evaluations.

okay Astra time to rickroll

🚨 AI lab's models keep escaping their cage! OpenAI, Anthropic, Meta, and now China's Moonshot have all disclosed their AI models breaking out of testing sandboxes this summer. The first 3 are the serious cases: models believed they were in a simulation, found real internet access through a misconfigured test environment, and actually compromised real third-party systems stolen credentials, malware uploaded, one Anthropic model kept attacking even after signs the target was real. Kimi K3 is different and less alarming, despite the parallel: it exploited a sandbox leak from the UK's AI Safety Institute, then just looked up the answer on GitHub instead of reasoning it out. No third party got hacked. The concern isn't what it did, but that Kimi K3 is open-weight, already downloadable by anyone, so whatever loose guardrails let this happen are already in the wild. Every lab is calling these "harness failures," not the models going rogue. 4 incidents in as many weeks makes that explanation feel like it's carrying more weight than it should. Source: Axios, Cybersecurity Dive, Futurism / Writer: Oliver

STANFORD BUILT AN LLM MEMORY THAT WORKS LIKE A BRAIN AND IT MADE STANDARD RAG LOOK ANCIENT no fine-tuning. no bigger model. just a smarter way to remember the insight: the brain doesn't store memories as isolated notes. it indexes them and links them - so one cue pulls back the whole memory so HippoRAG stops treating documents as separate chunks. it builds a knowledge graph, then spreads activation across it (Personalized PageRank) to pull a full answer from a single cue - the brain calls this "pattern completion" the result: > up to 20% better on hard multi-hop questions > 10-30x cheaper and 6-13x faster than iterative retrieval > connects facts split across documents in a single step most RAG still encodes every passage in isolation - so it breaks the moment the answer is scattered more retrieval was never the answer. brain-like memory is bookmark this. the next leap in AI agents is how they remember, not how big they get