A security researcher disclosed a pull-based prompt-injection attack called HalluSquatting that targets coding assistants and agentic tools including Cursor, Gemini CLI, Windsurf, and GitHub Copilot. …
In 2026, reinforcement-learning environments became the most contested resource in AI, with open-source lab Prime Intellect warning that big labs are locking them down and Anthropic reportedly plannin…
Prime Intellect launched Lab, a full-stack platform for training and optimizing AI models through agentic post-training, unifying environments, hosted training, and evaluations. The platform aims to d…
Prime Intellect raised $130 million in Series A funding led by Radical Ventures, with participation from NVIDIA Ventures, Intel Capital, and Dell Technologies Capital, to build an open superintelligen…
AI and deep-tech startups raised over $500 million in the past 24 hours, led by quantum-computing firm Oratomic's $300 million Series A and AI-agent startup Prime Intellect's $130 million Series A at …
Wired reporter Steven Levy built a self-improving AI using tools like AutoResearch and Prime Intellect, training small language models to automate newsletter tasks. The experiment shows that recursive…
Prime Intellect raised $130 million in Series A funding at a $1 billion valuation to provide enterprises with a full-stack platform for training and deploying their own AI agents, rather than renting …
Prime Intellect, a startup providing computing power and software tools for building AI agents, raised $130 million in Series A funding at a $1 billion valuation. The round was led by Radical Ventures…
Zenith, a new agent harness developed by an unnamed team, lifted OpenAI's GPT-5.5 from 5th to 1st place on the FrontierSWE benchmark, outperforming Anthropic's Claude Fable. The harness improves perfo…
Prime Intellect released prime-rl 0.6.0, an open framework for asynchronous reinforcement learning on trillion-parameter Mixture-of-Experts models, enabling training on agentic RL workloads with optim…
Microsoft, Hugging Face, Meta's PyTorch team, NVIDIA, and others launched OpenEnv, an open protocol for agent learning environments that standardizes how agents practice and improve. The protocol aims…
Anthropic CEO Dario Amodei said reinforcement learning shows the same log-linear scaling as pre-training, but RL system efficiency is critical to afford enough training. Experiments on open models sho…
CoinFund founder Jake Brukhman advocates for decentralized AI networks built on distributed GPU resources as a counterbalance to Big Tech's control over artificial intelligence. CoinFund has raised $1…
Anthropic released Claude Fable 5 on June 9 with built-in classifiers that silently downgrade users attempting to develop rival AI models, targeting Chinese AI labs. The company faced immediate backla…
Anthropic reversed its plan to silently degrade Claude Fable 5's performance for users training competing AI models, admitting the company "made the wrong tradeoff" after backlash from the research co…
A team of researchers from Yale's Building AI Infrastructure class achieved a new speed record of 81.2 seconds on the modded-nanoGPT benchmark by introducing a learnable per-(layer, head) tanh(α) gate…
Seven open-weight frontier models, including Hugging Face's SmolLM3, DeepSeek-R1, and OpenAI's gpt-oss-120b, were analyzed to distill common training methodologies for multi-billion parameter models, …