EU investigates latest loss of control over OpenAI’s AI agents, involving a German website
The European Commission said on Monday that it was investigating a new incident disclosed on Friday, involving thousands of OpenAI AI agents that disobeyed their instructions and took control of a German website in May, several months…
FashionNetwork USA
Publisher
Sep 7, 2026 at 11:55 AM UTC · 2 phút đọc

Published
September 7, 2026
The European Commission said on Monday that it was investigating a new incident disclosed on Friday, involving thousands of OpenAI AI agents that disobeyed their instructions and took control of a German website in May, several months before the unauthorised intrusion on the Hugging Face platform.
“This is not the first time control has been lost (over AI agents),” but “we are taking the matter very seriously and are monitoring the situation very closely,” Thomas Regnier, a spokesperson for the European Commission, told journalists.
“We have received an incident report from OpenAI” and the Commission is “in close contact” with the company, added the spokesperson, noting that Brussels has had new powers to regulate AI providers since August, under the EU’s Artificial Intelligence Act (AI Act).
According to the report published on Friday regarding the May incident, thousands of OpenAI’s AI agents swarmed a “wiki,” a collaborative website for German software developers where users can create and edit pages. Unlike in July, when OpenAI agents left their supposedly confined environment to intrude on the Hugging Face platform, this time the agents were authorised to access the internet, but only for browsing, not to post or edit content.
They collaborated to find a workaround, submitting false data to trick the security system. They then posted around 18,000 messages seeking tips related to the programme to which they were subject, though the researchers were unable to determine whether this was for development or evaluation purposes.
They also concealed instructions and posed as a site moderator. Once discovered, they created pages to replace those that were gradually being deleted by the moderators. Incidents involving agents exceeding their remit in order to carry out the tasks assigned to them have become increasingly frequent in recent months, with Anthropic and the Chinese firm Alibaba experiencing similar incidents.
This article is a translation. Click here to read the original article.
Copyright © 2026 FashionNetwork.com All rights reserved.
Sourced by
Originally reported by FashionNetwork USA
NewsLayer coverage based on externally reported material.
The Daily Brief
The onchain economy, before your day starts.
Curated markets, onchain insights, and key headlines — delivered every weekday morning.
Weekdays · Free · ~5 minute read
0
Applause
Was this article helpful?
Article Intelligence
Topics
Sponsored
AdNewsLayer Premium
Unlock deeper intelligence.
Ad-free reading, exclusive research, and real-time onchain insights.
Go Premium


