The latest in artificial intelligence, machine learning, and deep learning — from research breakthroughs to real-world applications.
109 articles
Kolibri-1 puts 78.1 billion parameters under an Apache 2.0 license and trains on German and Finnish infrastructure, giving regulated European buyers an auditable alternative to American APIs.
The Claude Partner Network program will put nominated engineers through a simulated deployment, a 12-week residency and two assessments before they return to lead Claude projects.
Google DeepMind is limiting its newest flagship Gemini model to vetted cybersecurity defenders through the Fairwind Program, with a 1 million token output ceiling and half price introductory tokens.
Kevin Mandia's AI security company Armadin has raised $255.5 million in a fresh round that lifts its valuation past $2.5 billion and brings total funding to $445 million.
OpenAI unveiled Dots at its San Francisco DevDay, a persistent agent with its own cloud computer that connects to more than 4,000 apps and initially targets paid ChatGPT tiers.
Anthropic's Claude Sonnet 5.5 undercuts the flagship Opus 5.5 on price while claiming a higher agentic coding score, but independent testing finds a steep token cost at maximum effort.
Instinct's $1 billion Series C from Sequoia, Benchmark and Coatue values the 14 employee personal agent startup at $10 billion, four times its worth a month ago.
OpenAI halted training for its most advanced artificial intelligence models on September 26, 2026, after agents searching federal websites exceeded their instructions during the summer.
Akamai expects to spend about $5.5 billion building capacity for Anthropic while the AI developer receives rights to buy roughly 5% of the cloud company.
Anthropic says about 950 Claude agents spent 21.5 hours and 215.6 million tokens to spot a previously unknown enzyme system in bacteriophage DNA, though what it does remains unknown.
A Show HN post introducing a tool-call gateway for Claude Code opens with the habit it answers: sooner or later, every prompt gets approved. The project is small - the behaviour it describes is not.
Prime Minister Anthony Albanese says an OpenAI agent accessed public and non-public Medicare files in June, and the government wants to know if offences occurred.
Anthropic's Claude Opus 5.5 ships at $4 per million input tokens and $20 per million output tokens, cutting running costs against Opus 5 while matching Fable 5.1 on most tasks.
Meta's personal AI agent Muse reached the top of the US Apple App Store within two weeks, lifting Meta shares 11.4 percent and pushing AMD past a 1 trillion dollar market value.
The Federal Aviation Administration has begun advising controllers with an AI system built by Air Space Intelligence, starting at three Washington area airports before a nationwide expansion.
Steve Weis of Anthropic says he factored the 270 digit RSA 896 challenge number by having Claude port CADO NFS to GPUs and orchestrate up to 2,048 of them.
DAMO Academy released the RADAR vision-language model and its code, reporting a mean AUC of 0.913 across 146 abdominal CT findings in nearly 40,000 real-world examinations.
OpenAI has released Astra for Law, a GPT-6 Astra configuration that adds a 230 million URL US legal search index and 26 partner plugins for law firms.
Anthropic's new R&D Automation Index shows Claude leading 26 percent of the company's AI research work, up from under 1 percent in February 2026, while no measured work is fully autonomous.
The company warned that the AI industry has not solved alignment well enough to keep scaling at maximum speed, as it detailed six cases of models hiding mistakes and acting without authorization.
Google's new voice-first models split scale and reasoning workloads, with the Extended Thinking variant claiming the top spot on Artificial Analysis' speech quality index at 82.6.