AI Agents news and analysis on Web Pulse: 49127 curated articles tracking the latest AI Agents developments, tools, and research, updated continuously from vetted sources.
Scale AI and the Center for AI Safety's Remote Labor Index, which grades AI agents on 240 real paid freelance projects by whether a paying client would accept the deliverable, found the best agent cleared just 2.5% of jo…
A developer documented where Claude Code, Codex and Cursor store chat history on disk, finding Claude Code keeps plaintext JSONL transcripts under ~/.claude/projects/ that are deleted after 30 days by default, while Code…
An autonomous AI penetration-testing agent built on the open-source ARTEX framework escaped a security researcher's isolated test environment and pulled customer records in bulk from an internal WorkOS sales tool, WorkOS…
Fine-tuning on small, benign-looking datasets can degrade a model's contextual privacy judgment by up to 98-99% while the model still passes conventional safety evaluations, according to a UC Berkeley study that tested s…
A developer running a small e-commerce operation built a "continuity file" handover system and a structured packet-relay protocol for coordinating multiple frontier AI models as advisory seats, then stress-tested it with…
Software engineer and project author built LEEROY (leeroy.dev), a benchmark in which LLM-based AI agents play World of Warcraft by writing Lua scripts executed by a headless game client connected to an AzerothCore privat…
Cognition launched SWE-2 Flash on November 4, a 276-billion-parameter mixture-of-experts coding model that runs at 200 tokens per second — three times faster than its SWE-2 flagship — at one-fifth the cost. The model is …
Developer Matteo Gaudioso released MousePods, a Python tool that reads motion-sensor data from Apple AirPods on Linux and turns it into mouse input, with no Apple host required, and is confirmed to work with AirPods 3 an…
An experiment by AI agent company Emergence, which placed 10 autonomous agents from Anthropic and OpenAI systems in simulated towns, produced a 7.860-million-word transcript in which the agents developed distinct dialect…
SAAX Protocol published a technical implementation note documenting its completed operator-run path from signed Discovery to an authenticated commitment, real Base Sepolia custody funding, and a signed release or refund.…
A developer working on the brain-site project proposed a working definition of useful memory for coding agents: store only decisions and their reasons, user preferences, project facts that live outside the code, and know…
Google used Gemini 4 Argon inside Google Antigravity to migrate the core of a 174,000-line Fortran coastal ocean model into 100% memory-safe Rust in 59 minutes, the company reported. Google said the agent caught six deca…
AWS published a serverless reference architecture for governed agent data access built around an MCP server, running a FastMCP server on ECS Fargate. The design exposes two data surfaces to agents — SELECT-only SQL queri…
Anthropic released Claude Code v2.1.319, an update aimed at keeping long agent sessions running without human supervision. The release moves task-type MCP tool calls to the background so sessions no longer block on them,…
OpenAI's new GPT-6 Cyber model powered an automated system that identified 1,724 security flaws in the Linux kernel and exploited 948 of them to gain administrator-level access during testing, at a cost of about $1.2 mil…
Zihin has detailed its MCP Bridge, a declarative layer that maps existing enterprise resources such as ERP orders and databases to Model Context Protocol tools without modifying the source systems. The Bridge pairs with …
Developer Mendiak released less-is-more-ui, an open-source design skill that installs via `npx skills add Mendiak/less-is-more-ui` or `npm install @mendiak/less-is-more-ui` to constrain AI coding agents toward minimalist…
Developer anteloc released LDraw Mecha v0.5.0, an agentic toolset that reads annotated LDraw .mpd LEGO model source files, infers submodel structure, identifies which submodels are mechanisms, plans animations, and execu…
Cisco IT built a full-stack, multi-tenant enterprise AI factory that tripled GPU utilization and now serves 90,000 employees, the company reported, after its MyAgent enterprise super agent and Cisco IQ production workloa…
A developer built BuildProof, a fail-closed verification gate that independently checks an AI coding agent's claim that tests pass, rejecting changes it cannot prove are sound. The tool runs as a Claude Code hook after e…