Agents - News Articles

Hacktron AI Claims Three-Person Team Breaches OpenAI Codebase in Under 72 Hours Using Claude-Assisted Attack

Sep 21, 2026

The Rundown AI

Hacktron AI says a three-person team breaches OpenAI’s private codebase in under 72 hours using a Claude-assisted attack, exploiting an image-upload flaw and staff sign-in tokens before leaving a signed internal-document edit and collecting a $6,500 bug bounty.

Security Agents

TypeSafe AI’s Jev System One Beats GPT-5.6 and Claude in LangChain’s Narrow Weather-Agent Evaluation

Sep 20, 2026

langchain

TypeSafe AI’s new Jev System One tops GPT-5.6 Luna, GPT-5.6 Terra and Claude Sonnet 4.6 in LangChain’s narrow weather-agent evaluation, matching human pass/fail labels on all 500 repeated decisions.

Foundation Models Agents Climate & Environment

Cua Launches Open-Source Platform for AI Agents to Operate Desktop Apps and Browsers

Sep 20, 2026

GitHub

Cua launches an open-source computer-use platform that lets AI agents operate native desktop apps and browsers across macOS, Windows and Linux through its Cua Driver CLI, MCP and typed SDKs.

Agents Coding & Developer Tools Open Source

Google Labs Launches CC, an AI Agent for Shared Family Planning and Household Tasks

Sep 18, 2026

Google

Google Labs launches CC, an experimental AI household agent that lets up to six U.S. family members share daily briefs, update calendars and tasks from emails, and complete permission slips or shopping lists through verified Google accounts.

Agents Consumer Products

Notion Launches Skills API to Bring Workspace Skills Into AI Agents and Tools

Sep 18, 2026

Notion

Notion launches its Skills API, letting teams turn Notion pages into AI-agent skills that sync to GitHub and can be installed through Vercel’s Skills CLI with commands such as `npx skills add <notion url>`.

Coding & Developer Tools Agents Enterprise

Agora launches Git-based shared memory for AutoResearch agents as 13 workers improve a frozen model without training or gradients

Sep 18, 2026

GitHub

Agora launches a Git-based shared-memory system that lets 13 AutoResearch agents improve a frozen 119.6M-parameter model without gradients or training data, producing 1,703 immutable DAG-recorded contributions and a reproducible 145-commit recipe.

Agents Research Foundation Models

Tamara Tran Launches Claude Code Plugin That Replaces Lossy Compaction Summaries With Jev-Scored Tool-Call Decisions

Sep 18, 2026

GitHub

Tamara Tran launches fast-jev-compaction for Claude Code, using Jev-scored decisions to preserve user and assistant text verbatim while pruning obsolete tool calls and truncating unneeded results to a default 300-character head.

Coding & Developer Tools Agents

Browser Use Launches Open-Source Jev Ultrafast Browser Agent, Cutting Google Flights Search Time to 7.1 Seconds

Sep 18, 2026

GitHub

Browser Use launches Jev Ultrafast as an open-source browser agent that completes a Zürich-to-London Google Flights search in 7.1 seconds, reducing median browser protocol calls from 1,092 to 101 through a dynamic indexed action space.

Agents Open Source Coding & Developer Tools

Anthropic Launches Claude Code Projects Beta for Parallel AI Coding Threads

Sep 18, 2026

Claude Code Docs

Anthropic launches Claude Code Projects in public beta, enabling eligible Pro and Max users to run parallel AI coding threads with shared repositories and memory, with up to 200 new threads daily that can open pull requests and fix CI failures.

Coding & Developer Tools Agents Cloud Computing

Perplexity Adds Effort Controls That Auto-Select Models and Coordinate AI Agents

Sep 18, 2026

Perplexity AI

Perplexity now adds Light, Standard, High, and Ultra effort controls to Computer, automatically selecting models and reasoning depth while higher settings use a coordinating AI to delegate work across agents from different providers.

Agents Consumer Products