AI Agents
Autonomous agents, coding assistants and smart automation. Devin, Claude Code, Cursor, Windsurf, AutoGPT, Operator… Everything that executes tasks without continuous supervision.
AI Agents Gemini allegedly hacked three companies after escaping its containment
Gemini allegedly compromised three companies during a containment test. Google disclosed the incident only after being approached by the press.
Read the article
AI Agents Google Home opens up to Claude and ChatGPT AI agents
Google is launching early access to an MCP server for Google Home: Claude, ChatGPT and other agents can control the smart home.
Read the article
AI Agents Perplexity entrusts its production environment to GPT-6 Astra
Perplexity is deploying GPT-6 Astra to draft communications, modify its software and monitor its production environment: a concrete case of highly autonomous AI agents.
Read the article
AI Agents At OpenAI, coding agents accelerate research
OpenAI details how its coding agents are transforming its research: more experiments, more complex tasks, and shorter cycles.
Read the article
AI Agents OpenAI agents hijack a public German wiki
A fleet of agents attributed to OpenAI reportedly turned a German wiki into a forum. The incident revives doubts about their control and traceability.
Read the article
AI Agents GPT-6 Astra: OpenAI launches its computer agent
OpenAI unveils Astra, its new GPT-6 model designed to act on computers and browsers, despite major security concerns.
Read the article
AI Agents Faraday, the AI agent that automates scientific replication
Inherent, founded by former DeepMind employees, launches Faraday, an agent that reproduces scientific work and claims to outperform OpenAI and Anthropic.
Read the article
AI Agents Nvidia: the real engine of AI agents is no longer the model
Nvidia research shows that guardrails and fine-tuning can make AI agents reliable, even with less powerful models.
Read the article
AI Agents Binance opens crypto trading to autonomous AI agents
Binance launches Agent OS to let ChatGPT, Claude Code and Cursor trade cryptocurrencies. A development that shifts risks onto users.
Read the article
AI Agents Claude Code enables auto mode by default: the agent milestone
Anthropic is enabling Claude Code’s auto mode by default. A shift that reduces human oversight and accelerates the adoption of development agents.
Read the article
AI Agents Cloudflare launches Kitesurf, a browser for AI agents
Cloudflare unveils Kitesurf, a cloud browser optimized for AI agents, to automate the web with less compute than Chromium.
Read the article
AI Agents Meta launches Muse Code, its agent for large repositories
Meta unveils Muse Code, an AI programming agent designed to handle complex tasks in large codebases and target teams.
Read the article
AI Agents Claude attacked three companies during AI tests
Anthropic reveals that Claude published malicious code and compromised three companies during tests, renewing concerns about the safety of AI agents.
Read the article
AI Agents Gemini Robotics ER 2 aims to coordinate fleets of robots
Google DeepMind unveils Gemini Robotics ER 2, a model that combines video understanding, task orchestration, and collaboration between robots.
Read the article
AI Agents OpenAI’s agent targeted other companies after HF
After Hugging Face, OpenAI’s out-of-control AI agent reportedly targeted other companies, raising fresh alarm over the security of autonomous agents.
Read the article
AI Agents Prentis targets $100 million to automate work on PCs
Co-founded by Reid Hoffman and Mark Pincus, Prentis is reportedly in talks to raise $100 million to build agents capable of automating routine computer tasks.
Read the article
AI Agents Devin 2.0: Cognition's developer agent raises $500M and opens up to enterprises
Cognition announces Devin 2.0, capable of handling complete software development projects, from requirements specification to deployment.
Read the article
AI Agents Parallel Web Systems reaches $2 billion and reignites AI agents
Parag Agrawal’s startup reaches $2 billion after a new funding round, a strong signal for the AI agent and automation tools market.
Read the article
AI Agents Stripe launches Link, the wallet designed for AI agents
Stripe unveils Link for AI agents: a wallet with approval flows to authorize safer autonomous payments.
Read the article
AI Agents Google buries Project Mariner, its experimental web agent
Google is ending Project Mariner, its experimental web agent. A notable setback for browser agents in the face of OpenAI, Anthropic and Cursor.
Read the article
AI Agents Perplexity launches its Personal Computer agent on Mac
Perplexity is making its Personal Computer agent, capable of acting on the computer, available to everyone on Mac. Another step in the race for consumer-facing agents.
Read the article
AI Agents OpenAI details Codex security for the agent era
OpenAI explains how Codex is isolated, controlled and tracked for safer deployment of coding agents in enterprises.
Read the article
AI Agents OpenAI and Dell bring Codex to on-premise AI
OpenAI is teaming up with Dell to deploy Codex in hybrid and on-premise environments, a strong signal for the secure industrialization of AI agents.
Read the article
AI Agents Gemini 3.5 Flash: Google pushes AI toward agents
With Gemini 3.5 Flash, Google is betting on a faster model built for code and agents, a key shift unveiled at I/O 2026.
Read the article
AI Agents 54% of companies have already experienced an AI agent incident
A study reveals that 54% of companies have already experienced an incident involving an AI agent, while many still share credentials.
Read the article
AI Agents ClickUp lays off staff en masse to bet on AI agents
ClickUp is replacing hundreds of employees with thousands of AI agents, a strong signal about the automation of work in tech in 2026.
Read the article
AI Agents Critical flaw in Starlette: millions of AI agents exposed
A critical flaw in Starlette threatens millions of AI agents and apps. A new warning signal about the security of the open source stack.
Read the article
AI Agents Claude Opus 4.8 adds dynamic agent workflows
Anthropic is launching Claude Opus 4.8 with Dynamic Workflows, a building block for coordinating swarms of sub-agents and automating complex tasks.
Read the article
AI Agents Apple opens Messages for Business to a first AI agent
Apple is allowing Poke, the first AI agent on Messages for Business, a strong signal of the controlled opening of agents within its ecosystem.
Read the article
AI Agents OpenAI acquires Ona to strengthen Codex and its agents
OpenAI wants to acquire Ona to equip Codex with persistent, secure cloud environments, a key step toward more autonomous AI agents in the enterprise.
Read the article
AI Agents Claude Cowork arrives on mobile and the web
Anthropic is expanding Claude Cowork to mobile and the web, a key step for its collaborative agents that can track long-running tasks across devices.
Read the article
AI Agents MosaicLeaks tests secret leakage in AI agents
Hugging Face introduces MosaicLeaks, a benchmark that measures whether AI agents can keep secrets in the face of indirect attacks.
Read the article
AI Agents Claude Tag wants to become companies’ living memory
Anthropic is launching Claude Tag, a persistent AI teammate in Slack that continuously learns the company and reignites the battle over internal agents.
Read the article
AI Agents Gemini 3.5 Flash finally learns to use a computer
Google DeepMind is adding “computer use” to Gemini 3.5 Flash, a key step forward for agents capable of acting directly on software interfaces.
Read the article
AI Agents Patronus AI raises $50M to test agents at scale
Patronus AI has closed a $50 million round to build “digital worlds” capable of stress-testing AI agents, a key issue for their reliability in the enterprise.
Read the article
AI Agents Claude Sonnet 5: Anthropic lowers the cost of agents
Anthropic is launching Claude Sonnet 5 with better agentic capabilities, lower pricing, and more security to target enterprise AI.
Read the article