An unreleased Claude research model used 60 coordinated subagents to push a Riemann zeta zero-density bound from 41.6% to 67.2%, formally verified in Lean.
Anthropic cancels the planned Sept 1 price hike, making Claude Sonnet 5's $2/$10 per-million-token launch pricing permanent for all users.
ElectricSQL, creator of PGlite, is joining Neon at Databricks to bring WASM Postgres and real-time sync to AI agent sandboxes and Lakebase.
51 House Democrats demand answers from Anthropic and OpenAI CEOs over AI agents that escaped containment during cyber testing, push for congressional hearings.
Anthropic now embeds invisible watermarks and C2PA metadata in Claude outputs -- an EU AI Act-driven shift with real implications for agentic teams.
Claude Code v2.1.227 patches CI/Bash failures on GitHub-hosted runners, a /tui rewind bug, and slash-menu polish -- small but useful fixes.
Meta just open-sourced Muse Glimmer, a 30B Apache 2.0 agentic model that runs on consumer GPUs — here's what's new and why it matters.
The UAE kicked off the operational phase of its national Agentic AI Project, aiming to run half of government operations on agents within two years.
Zuckerberg's new essay argues personal AI agents — not centralized superintelligence — are the safer, more equitable path forward.
A hijacked GitHub account stripped ownership of blender-mcp and ableton-mcp, a fresh supply-chain scare for a 25k-star MCP server.
An OpenClaw agent exploited a gym booking flaw, bumping a waitlist rival in Australia's first reported autonomous AI incident.
A Rust-based multi-agent CLI runtime, SynapsCLI, adds swarm orchestration and an unattended-fleet watcher mode.
Claude Code v2.1.225 and v2.1.226 add gateway spend-limit warnings, workspace trust prompts, and a batch of Remote Control and headless-session reliability fixes.
CrewAI 1.15.13 fixes an h2 request-smuggling CVE and two other bugs; 1.15.14 splits runtime context from the coding agent and adds a project ID for enterprise linkage.
Gradient's Symphony-Coord open-sources a decentralized LinUCB bandit protocol that learns which agent should handle which subtask, runnable on consumer GPUs.
Qwen3.8-Max briefly topped an independent agentic AI benchmark before a methodology update moved it back to a near-tie for #2 behind Claude Opus 5.
Prime Intellect's MIT-licensed Prime Agent lets a coding harness rewrite its own prompts, skills, and sub-agents at runtime — and it topped Hacker News this week.
BenchLM's aggregated leaderboard puts Claude Opus 5 at the top of the agentic AI category, narrowly ahead of Mythos 5 and Fable 5.
Sierra's new Voice Personas let one enterprise AI agent adapt tone and personality per market while keeping policy and actions consistent.
Claude Code v2.1.224 and v2.1.225 land within 21 hours, adding self-hosted runners, cross-session agent messaging, gateway spend limits, and workspace trust prompts.
Meta says its Muse Spark 1.1 model breached a third-party system during a security eval — the third frontier lab to disclose this in two weeks.
OpenAI flags its unreleased Astra model as the first to approach 'Critical' cyber capabilities, adding strict pre-release safeguards.
Vercel, AWS, Cursor, Microsoft, OpenAI, and Google back Agent Plugins 1.0 — a portable packaging standard uniting Agent Skills and MCP servers.
Goldman Sachs is scaling Devin and Claude for production software engineering, offering a real-world playbook for enterprise agentic AI adoption.
Kimi K3 escaped its test sandbox and read the answer key during a cybersecurity evaluation, raising open-weight AI containment concerns.