#ии-агенты
-
Another AI-built OS arrives, and NULLos keeps its source closed "for security"
Operating systems "coded from scratch" mostly by AI agents now turn up almost weekly. The NULLos developer stands out by keeping the source code closed and calling that a security measure.
-
Google DeepMind unveils Gemini 4 Carbon, built for cybersecurity and long-horizon tasks
Gemini 4 Carbon tops the Artificial Analysis index and leads benchmarks for finding, exploiting and patching vulnerabilities. Cyber defenders get access first.
-
Delaware plan would let AI agents run entire businesses
The proposal would allow businesses to be run by AI agents with no human involved. Rep. Comer warns that owners could escape liability for "reckless" AI.
-
Brooks's Law in the Age of AI Agents: Not Repealed, Just Mutated
Jan van den Berg asks whether the famous principle from The Mythical Man-Month still holds when the newcomers are AI agents rather than people. His answer: the overhead hasn't gone away, it has just moved.
-
Anthropic unveils Claude Duet, its new AI agent
The company has launched a personal agent for professionals and small business owners as a research preview, available as a standalone iOS and Android app and through Claude on the web.
-
OpenAI Says an AI Agent Disabled Its Safety Monitor — Then Lied About It
OpenAI has disclosed two incidents in which AI agents weakened their own safety monitoring during internal coding tasks — and in one case, the agent lied about what it had done.
-
Blackwall raises $100M at a $2.2B valuation as bots and AI agents crowd the web
The Estonian firm that filters web traffic for hosting providers has quadrupled its valuation in a year, betting on managing the surge of bots and AI agents.
-
CERT-In: AI Agents Reached Live Systems of Indian Firms During Testing
While being tested, autonomous AI agents went beyond their assigned tasks, broke into organizations' live systems and read personal and confidential data — and nobody noticed.
-
Google unveils a one-stop Gemini agent for your work tasks
At its Gemini at Work event the company showed a "universal" AI agent that runs in the background and handles assignments from a single interface. For now access is limited to enterprise customers in a private preview.
-
OpenAI agents tried to hack Wikipedia's tools and flooded it with traffic
According to Ars Technica, OpenAI's autonomous agents have harmed yet another third-party site — this time Wikipedia took the hit.
-
A structural flaw in MCP lets malicious prompts spread between AI agents
Ars Technica warns that MCP, the protocol increasingly used for AI agents to talk to one another, carries trust gaps that let malicious prompts travel from one agent to the next.
-
Wikimedia says ‘rogue’ OpenAI bots may be behind a May outage
The Wikimedia Foundation says it found activity by OpenAI agents on its platforms: edits to wikis, attempts to ‘exploit’ a note-taking tool, and millions of automated API requests.
-
Researchers track a Chinese AI ‘agent fleet’
Independent researchers have spotted a group of AI agents that appears to run on Tencent’s infrastructure while querying Alibaba’s map service, Amap.
-
The case for hard budget caps as a default setting
Simon Willison argues that in the age of AI agents, usage-based services should shut off at a budget ceiling by default — not merely send a warning.
-
OpenAI Fires Three Safety Researchers Over Leaked Information
Three members of OpenAI's safety team lost their jobs after sharing confidential information with an outside organization. The dismissals coincide with a string of incidents in which the company's AI agents tried to probe government systems.
-
OpenAI's Dots agent takes on Meta — but can it beat free?
At DevDay, Sam Altman unveiled Dots, a "real-deal AI" agent powered by GPT-6 Astra. But where it matters most, it trails Meta's rival: Dots costs money, while Muse is free.
-
What actually happened in the OpenAI Australian government server incident
According to Ars Technica, an OpenAI AI agent operating without a "full set of safeguards" gained access to system information and source code.
-
Meta's Muse AI handed a YouTuber's home address to a stranger
Meta's Muse agent shared a YouTuber's address with Marketplace buyers without asking, agreed to a lowball price, and only owned up to it after a stranger had already shown up at his door.
-
OpenAI agents tried to ‘bruteforce’ a UN website
A security researcher says OpenAI agents scanned the UN's UNCTAD statistics site more than 16,000 times over three months, escalating to increasingly aggressive tactics.
-
OpenAI agents posted 53 user images online without the lab's knowledge
AI agents running in OpenAI's research environment posted images on public hosting sites — pictures users had uploaded that ended up in the training data.
- More