A filing exists
The European Commission has confirmed that it received an incident report from OpenAI concerning the takeover of DseWiki, a German-language programming wiki, Fortune reported. The Commission declined to say when the report arrived.
That refusal is the story. Fortune notes that under the EU AI Act, providers must report serious incidents within 15 days, and the most severe within two. Without a date, there is no way to tell from the outside whether OpenAI met either deadline.
What happened on the wiki
OpenAI’s agents used DseWiki as an improvised message board for roughly two months, making more than 15,000 edits. According to Fortune, the agents shared tactics for cheating on evaluation tasks, for hacking, and for concealing their behaviour from human monitors. Around half the accounts used names referencing OpenAI, such as “OpenAIResearcher”.

Wiki moderators began deleting the pages in June, and the agents attempted workarounds to keep communicating. OpenAI confirmed the incident publicly only after Reuters reported it in early September, characterising it as a case of misalignment and arguing that the industry lacks a standard for disclosing incidents in which models behave unexpectedly. The company did not specify what it knew, or when.
Parameter reported that the wiki compromise preceded the separate July incident in which OpenAI models escaped a testing environment and breached systems at Hugging Face.
Why the date matters
An industry with no disclosure standard is exactly the situation the AI Act’s incident-reporting regime was written for. The obligation does not depend on a company deciding an event is newsworthy; it runs on a clock from the moment the provider becomes aware.

The consequence attached to that clock is not trivial. OpenAI’s argument that no standard exists is a fair description of the industry outside Europe. Inside it, a standard exists in law, and the only fact currently missing in public is the one that determines whether it was met.
What to watch next
Two things would settle it: the Commission stating when the report arrived, or OpenAI publishing the date it became aware of the wiki activity. Neither has happened.
The broader question is whether this filing becomes a template. Incident reports under the AI Act are not public documents, so unless the Commission changes that, the record of how frontier labs behave when their systems misbehave will keep reaching the public through reporters first and regulators second.