Microsoft released MAI-Cyber-1-Flash, its first dedicated AI security model, and MDASH, a new agentic cybersecurity platform to automate enterprise security operations, in direct response to the OpenA…
Cybersecurity researchers report that safety guardrails from OpenAI and Anthropic are preventing legitimate offensive security work to find unknown vulnerabilities and develop exploit tools, according…
Cognition acquired Poke, a conversational AI startup, in a deal valuing the startup in the low nine figures, according to TechCrunch. The acquisition will bring Poke's interaction model and personalit…
Prentis, an AI lab co-founded by Reid Hoffman and Mark Pincus, is in talks to raise $100 million at a $1 billion valuation to develop computer use models that automate routine office tasks, according …
Multiple new tools, including TerminAI and Termic, launched to integrate AI agents directly into command-line interfaces, enabling Claude and Codex to analyze shell output and compose commands without…
Developers running agentic AI coding tasks report significant idle time during autonomous execution and difficulty managing parallel agent sessions across Git worktrees, according to Hacker News discu…
New tools Hubo and frameworks A2A/AP2 implement peer-review and incentive-based patterns for multi-agent code collaboration, according to GitHub repositories and engineering blogs. Hubo uses two agent…
InfoTrack has built a dedicated MCP business unit to serve AI agents as customers, deploying an enterprise MCP gateway with a registry and search-and-invoke pattern. Ajay Kumar explains that human-in-…
Nearly two-thirds of all tracked AI job postings specifically build or run agents, according to Get Ready For Agents' live indexes covering 216 companies actively hiring for agent roles from a tracked…
Five major large language models (LLMs) — Claude, Codex, Gemini, and two others — consistently hallucinate the same 127 nonexistent Python and npm package names, according to researcher Aleksandr Chur…
JetBrains benchmarked Rust Token Killer (RTK), a compression proxy for Claude Code, and found it increased token costs by 7.6% at low reasoning effort (p=0.004) with no measurable change at high effor…
Anthropic released Claude Opus 5, which ranks first on the Artificial Analysis Intelligence Leaderboard and achieves near-Fable 5 capabilities at lower cost by removing over 80 percent of system promp…
Zenity Labs discovered AgentForger, a phishing-based attack that silently creates and launches fully autonomous AI agents within OpenAI workspaces with access to Outlook, Slack, SharePoint, and Google…
Screenpipe, a Y Combinator S26 company, has launched a tool that records screen and audio locally to create searchable memory for AI agents, enabling automation of repetitive tasks based on observed w…
Uber exhausted its entire annual AI infrastructure budget in just four months running Claude Code agents due to minimal cost controls and lack of token-to-outcome linkage, according to CIO.com. The in…
Codeberg e.V., the Berlin-based non-profit behind the open-source code hosting service, voted to ban 'vibe-coded' (AI-generated) projects and prohibit the use of user code or data for AI training, cit…
OneCLI released an open-source credential vault that prevents AI agents from leaking secrets during task execution, and Screenpipe (Y Combinator S26) launched a tool that records screen and audio loca…
Developers released browser-bridge MCP for logged-in Chrome automation and serve-avd for streaming Android emulators to browsers, extending agent capabilities beyond command-line interfaces, according…
5dive launched an open-source agent orchestration framework that runs multiple named agents (Claude, Codex, Pi) on user-owned servers, enabling organizational structures with shared backlogs and task …
OtoDock released an open-source platform that lets users run Claude Code and Codex agents as coordinated teams on self-hosted servers, supporting up to 5 users without requiring a credit card and inte…