AI Agents news and analysis on Web Pulse: 37109 curated articles tracking the latest AI Agents developments, tools, and research, updated continuously from vetted sources.
TypeSafe AI announced Jev on 15 September 2026, a "System One model" that takes a piece of state and a list of typed questions and returns typed answers with calibrated probabilities in a claimed 70 to 500 milliseconds w…
Tencent has open-sourced BrowserSkill (bsk), a lightweight local bridge that connects shell-capable AI coding agents such as Claude Code, Cursor, and Codex directly to a developer's already logged-in browser. The tool us…
Suprmind, a platform founded by Radomir Basta, runs frontier AI models side by side in a single shared thread so they can read and challenge each other's answers, a setup Basta says filters out false data and flawed reas…
Europe has live end-to-end agentic payments running across more than 30 banks, with Santander, Mastercard, ING, and Worldline executing transactions and ING, Worldline, and Visa completing an agentic payment in Germany o…
Five enterprise vendors shipped MCP-integrated governance products in the last two weeks, with Rubrik, Google, ServiceNow, Cisco, WSO2, and NVIDIA each staking a different tier of the agent governance stack rather than c…
Developer Vidit Raj released Pebble, an open-source 25-million-parameter language model built entirely from scratch that can call tools such as a calculator or weather API, publishing the code on GitHub. The build walks …
Google introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, two live-dialogue AI models that can hold real-time conversations while handling visual context and executing background tasks. The standard model …
A false AI-generated intelligence report nearly caused the U.S. military to intercept and board a Chinese ship in the Middle East during the war with Iran, CNN reported Friday, citing four people familiar with the episod…
A developer reports that OpenAI's Codex agent spent six days working on the goose codebase without completing a single run, while GPT-5.6 Sol driven from goose's own engine performed well on the same repository in the sa…
A University of Wyoming Department of Transportation production deployment found RAG accuracy fell from 75% to below 40% as its corpus grew from 54 to 1,128 documents (88,907 chunks), a failure the researchers formalize …
Timedomain released ACE Studio version 2.1.8, an update to its AI-powered music production workstation and virtual vocal synthesizer that adds VST instrument loading and CLI/MCP support. The CLI/MCP support lets users di…
TypeSafe released an interactive multi-drone autonomy simulation in which its Jev System 1 reflex layer controls a configurable fleet of one to twenty simulated drones, with an optional System 2 planner powered by OpenRo…
A developer rebuilt a Forge manifest major-version predictor after discovering their original tool was built on a misattribution: the 3.x to 4.0.0 bump of Sentinel Vault was triggered by both the new llm module and a new…
OpenAI researcher Noam Brown said on the Dwarkesh Podcast that air-gapping computers may not stop a misaligned AI, citing academic studies in which two isolated machines communicate via CPU heat and temperature sensors. …
Mastercard announced a partnership with Alchemy to launch a consumer-facing agentic payments option that gives AI agents a virtual Mastercard with user-set spending caps, merchant restrictions, and optional pre-checkout …
Anthropic reported on September 17 that its Claude model now leads 26% of the company's AI research and development work, up from essentially zero in February, with Claude authoring more than 80% of code merged into Anth…
Iran and China have created autonomous A.I. influence campaigns, according to a New York Times report published September 18, 2026. The story, which drew 3 points and 1 comment on Hacker News, describes the two countries…
Cayu released a production agent runtime for building and operating AI agents in Python, according to its GitHub repository. The framework gives applications full control over context assembly, model invocation, tool acc…
TurnPanel launched a workspace platform that unifies multiple AI models, agent conversations, browser activity, files, memory, and tools in a single interface. The platform lets agents pause before execution steps, givin…
A survey reported by CIO.com found that 52 percent of CIOs say they are blamed for mistakes made by autonomous AI agents even when they lack visibility into agent actions or control over vendor selection. The finding, dr…