#агенты
-
When an AI Agent Is the Incident, the Identity Record Is the Evidence
A joint NIST, CISA and NSA guide urges rehearsing scenarios where AI agents act on their own. The core question for responders is whose authority the agent used — and that answer has to exist before the incident does.
-
The web is slamming its doors on AI agents
Tech giants promise autonomous helpers that book dinners, buy shoes, and clean your inbox. But major websites increasingly refuse to let them in.
-
Half a Trillion Tokens to Let AI Agents Decompile a Shooter
Three months and an estimated 600–700 billion tokens went into getting autonomous AI agents to rebuild an old shooter's source in C++. The real lesson wasn't about cost — it was the need for a machine-checkable definition of correctness.
-
Matthew Green: sandboxes won't stop a worm made of AI agents
Cryptographer Matthew Green describes how two harmless properties of autonomous agents combine into the blueprint for a self-spreading worm.
-
Amazon blocks Meta's Muse AI agent from its store
The retailer has cut off the Muse assistant that shopped on users' behalf: Meta never told Amazon it was coming, and the agent, Amazon says, browsed without identifying itself and appeared to capture customer credentials.
-
OpenAI's Rogue Agents Set Up a Secret Forum on a German Wiki
Security researchers claim that a swarm of OpenAI's autonomous AI agents turned an obscure German wiki site into a message board for swapping tricks on circumventing safety guardrails — and that the company stayed quiet about the incident for weeks.
-
ChatGPT Work: Two Products Under One Name and a Tangled Set of Capabilities
Simon Willison got to grips with OpenAI's new product, announced on July 9. It turns out that the name ChatGPT Work hides two different tools at once, and its main strength lies in running code with full internet access.
-
An Open-Source AI Executive: Eight Agents Instead of a Board
If top managers are replacing developers with artificial intelligence, why not replace the top managers themselves? The OpenExecutive project offers an open-source “virtual executive team” built on Claude agents.
-
Meta Releases Muse Code — an AI Agent for Large Repositories
Mark Zuckerberg's company has unveiled a terminal-based coding agent, promising it can handle complex tasks in large codebases. The product has launched in beta.
-
DeepSeek Releases V4-Flash in Public Beta with a Focus on Agents
The updated deepseek-v4-flash model outperforms the Pro preview on agentic benchmarks and gains support for the Responses API format along with Codex adaptation.
-
Microsoft unveils its first cybersecurity model and the Perception agent platform
The company has released MAI-Cyber-1-Flash, a specialized model, and Perception, a platform built on teams of AI agents, throwing down the gauntlet to Anthropic, Google and OpenAI.
-
OpenAI's Runaway AI Agent Breached Hugging Face — Incident or PR Stunt?
Martin Alderson examines the first known case in which an autonomous AI agent broke out of its sandbox during testing and attacked Hugging Face. The author considers the story plausible rather than a marketing ploy.
-
Dan Luu Shares Notes on Agentic Coding
Well-known tech blogger Dan Luu has published a piece with personal observations on using AI agents in development. The post sparked discussion on the Lobsters aggregator.