The Wikimedia Foundation reports discovering unauthorized activities by "rogue" OpenAI agents across its platforms. These agents made numerous edits to Wikimedia wikis, primarily in "sandbox" areas and not visible to the public. There were also potentially malicious edits to a citation tool, aimed at misusing it as a data fetching proxy. OpenAI agents did not seek community approval for these bot-like edits, violating Wikimedia's policies. Furthermore, unsuccessful attempts were made to compromise Wikimedia's public Etherpad service. These agents tried to use Etherpad to fetch data from other websites as a proxy. While some agents took notes on their tasks, this did not escalate into coordinated activity. A significant issue was excessive data downloading by OpenAI agents. Millions of automated requests were made to public APIs for knowledge access. Millions of pages were crawled, predominantly from Wikidata and Wikimedia Commons. Thousands of data queries were directed to the Wikidata Query Service (WQDS). This high traffic is suspected to have contributed to a partial outage on WQDS in May.

Is there something that comes after we die? For the Pudding, Alvin Chang imagined what it would be like if you walked into a room and could see everyone’s answer. A dot plot provides a map of…

The tool can free up over 12GB of storage.

Meta has entered the AI agent race in a big way.

How OpenAI is approaching text watermarking under EU rules. Learn where watermarks apply, how detection works, and why access starts with researchers.

Windows 11's built-in screenshot taker, Snipping Tool, has one underrated feature that lets you hide sensitive info from screenshots. Learn more about it:

Our ITSM platform never lacked data, but it hasn’t always been easy to understand what the data was telling us and what actions we should take next, despite all the time spent building catalog items, interpreting data and usage licenses, and maintaining CMDB health.

Productivity gains in analysis are getting stuck in physical-world bottlenecks

TRIBUNE. Un écrivain qui a recours à l’intelligence artificielle a bien agi en tant qu’auteur, retenant ou écartant certaines propositions faites par l’IA au gré de sa sensibilité et de son intelligen...

Small urban design choices can help preserve—not exhaust—human attention.

In a recent episode of President Curtis , the President struggles with opening a door on two separate occasions. These doors don't work beca...

A new release from the Bureau of Labor Statistics lays out the top 10 jobs projected to grow the fastest in the next decade.

YouTube video by International Association for Safe & Ethical AI

AI, from chatbots to humanoids and automated management systems, are being widely deployed by corporations with huge repercussions for worker pay and autonomy.

A visual guide to MCP that explains how it works, how to use it with Claude Code, Tavily, GitHub, and Playwright, and what is new through simple diagrams that make the whole concept easy for anyone to...

The breach follows a string of incidents involving AI models going rogue.