Researchers from MIT and Sakana AI published SIFT (Self-Improvement via Fast Tree-search), a framework that uses a large language model as a pairwise referee to cut the cost of evaluating self-improvi…
Sakana AI's SAIL method raised simulated robot task success from 25% with one search node to 73% at 45 nodes by using Monte Carlo tree search over candidate trajectories evaluated in simulation, accor…
Sakana AI and the University of Tokyo introduced SAIL (Scaling In-Context Imitation Learning), a method that raises the average success rate of VLM-generated robot trajectories from 25% to 73% by scal…
Sakana AI researchers Jeffrey Seely and Julian Gould introduced Augmented Lagrangian Predictive Coding (PC-ALM), a layer-local alternative to backpropagation that attaches a Lagrange multiplier to eac…
Sakana AI released Fugu Ultra v2, a higher-performance model in its Fugu family that functions as a learned multi-agent orchestration system rather than a single monolithic model. Fugu Ultra v2 routes…
Sakana AI released two orchestration models, Fugu Max and Fugu Ultra v2, on September 10–11, with Fugu Max priced at $2 per million input tokens and $6 per million output tokens — 40 to 60 percent bel…
Saudi Arabia's HUMAIN unveiled the Arabic-language model humain-m3, commissioned from MiniMax, on September 3, and France's Mistral AI raised €3 billion at a valuation exceeding €21 billion, developme…
OpenAI has turned the Codex harness into an API, launching the Agents API in public beta with no platform fee, charging only for models and tools. A single API call can launch an agent with its own co…
Sakana AI released Fugu Max and Fugu Ultra v2, two new models in its Sakana Fugu family, live today through its OpenAI-compatible API with no open weights and no EU/EEA availability. Fugu Max is price…
Sakana AI released Fugu Max and Fugu Ultra v2, two configurations of its Fugu orchestration architecture that route tasks across a pool of open and specialized models, including the NVIDIA Nemotron fa…
Prime Minister Lawrence Wong outlined a philosophy for Singapore's technology approach at the National Day Rally on Aug 23, emphasizing harnessing technology on the country's own terms. The article pr…
Japan's Ministry of Internal Affairs and Communications now requires AI firms, including OpenAI, Anthropic, Google, Preferred Networks, and Sakana AI, to disclose training data sources, with violation…
OpenAI has halted reinforcement learning for its next-generation Astra model to strengthen safety measures, marking a shift from policy debates to engineering constraints in AI development. The pause,…
A team from the University of Cambridge's Machine Learning Systems Lab, working with NVIDIA, Flower Labs, MBZUAI, and Inria, introduced the Red Queen Gödel Machine, a co-evolutionary system that impro…
A Cambridge-led team with collaborators at NVIDIA, Flower Labs, MBZUAI, and Inria introduced the Red Queen Gödel Machine (RQGM), a framework that co-evolves an AI agent and its evaluator to prevent se…
Anthropic is rolling out an invisible watermarking system for its Claude AI model worldwide, using Google DeepMind's SynthID-Text protocol to comply with the EU AI Act, a move that sets the strictest …
A study posted on arXiv in late July found that AI systems are not yet capable of fully automating open-ended research, with an AI system built by Princeton University researchers scoring 2/6 and 1/6 …
Google's new ScientistOne framework recorded zero hallucinated references across 75 evaluated papers and 337 individual reference checks, compared to baseline systems with hallucination rates as high …
Japan's government is considering incorporating domestically developed AI into the Self-Defense Forces' command and control system for the first time, aiming to speed up decision-making and maintain A…
Four former Google researchers — Jeff Dean, Sanjay Ghemawat, Quoc Le, and Oriol Vinyals — left Google in August 2026 to launch Discovery Loop, a company betting that the fastest route to superintellig…