Every article we have published, newest first. Search titles and leads, or change the order.
BusinessNews
OpenAI is close to raising up to $30 billion at a $1.25 trillion pre-money valuation, led by UAE investors and Coatue, as a bridge ahead of a potential IPO, according to CNBC's sources.
· 2 min read
ResearchNews
Anthropic and Theorem used Claude to prove at machine-code level that 239 fuzz entry points never crash on inputs under 64 MB, finding long-hidden bugs along the way.
· 3 min read
ResearchNews
Two outside researchers spent three weeks inside OpenAI trying to break the system that watches its internal agents. They found several new vulnerabilities, including one that could hide hundreds of agent actions.
· 3 min read
ResearchNews
Microsoft AI published its MAI-DxO diagnostic system in Science. On hard NEJM cases it reached 83% accuracy, versus 36% for physicians using books and the internet.
· 2 min read
ModelsNews
Hugging Face released Carbon-X, a 14.8B-parameter Apache 2.0 model. You paste raw DNA into a chat and it answers in English, for example about a mutation's likely effect.
· 2 min read
ToolsNews
1Password's new platform lets personal AI agents sign in to your accounts without ever seeing or storing the password. Claude Duet and ChatGPT Work are among the partners.
· 2 min read
ModelsNews
iFlytek unveiled Spark X2.6 on October 22: a 293B-parameter MoE model with a 1-million-token context, trained entirely on domestic Chinese chips according to the company. A 4B on-device sibling is open under Apache 2.0.
· 2 min read
ToolsNews
OpenAI made the Decisions API generally available. Built on GPT-6 Luna, it answers bounded questions with a probability, a choice or a score, up to 10x faster.
· 2 min read
ToolsNews
Anthropic is handing the Agent Skills open standard to the Agentic AI Foundation (AAIF) under the Linux Foundation. SKILL.md based skills already run on more than 70 agent platforms, including ChatGPT, Cursor and GitHub Copilot.
· 2 min read
ResearchNews
Ai2 released OlmoEarth v1.3, a satellite embedding model, together with precomputed 10 meter annual embeddings for the whole planet and a browser-based explorer. Weights, training data and code are all open.
· 3 min read
ResearchNews
Microsoft open-sourced ThinkingBox-Training. It grades an agent by the final database state and uses that same verdict as the reward for reinforcement learning.
· 2 min read
BusinessNews
xAI launched the SpaceX Super Intelligence Partner Network, its official program for companies building products and services on Grok. More than 130 companies took part in the early-access phase.
· 2 min read
ToolsNews
At GTC Berlin NVIDIA launched Minerva, an AIOps platform that unifies infrastructure and workload telemetry for AI factories. Purpose-built AI models filter alerts, predict failures and point to the likely root cause.
· 2 min read
ToolsNews
Anthropic has opened a research preview of Claude Duet, a standalone app where Claude reaches out when something needs you and keeps working in the background on its own machine.
· 2 min read
ModelsNews
Baseten's research team, Base Labs, released two open decision models: the 27B mammoth-1 tops the Jev Decision Index, and the 9B elephant-1 leads models its size.
· 2 min read
ModelsNews
Gradium released Phonon, a text-to-speech family whose smallest 31M-parameter model is 64 MB, Apache 2.0 licensed and runs in a browser, on a phone or on a laptop CPU.
· 2 min read
ResearchNews
Letta's new benchmark does not test what an agent remembers but whether memory written from past runs helps solve new tasks. Opus 5.5 and GLM-5.3 lead at 73.3 percent.
· 2 min read
ResearchNews
Xingjian Bai and colleagues built an optimizer where a coding agent watches training and rewrites the update rule on the fly. Without a tuned recipe, it matches or beats off-the-shelf ones.
· 2 min read
ToolsNews
Breken rebuilds your software on its own machines, lets synthetic users loose on it and injects faults. When a failure reproduces reliably, it opens a draft fix PR.
· 2 min read
ResearchNews
Artificial Analysis has a new benchmark that measures the communication style of 38 AI models across 30 personality traits, sycophancy included. GPT-4o flatters most, Grok 4.7 least.
· 2 min read
