OpenAI agents accused of unauthorized activity on Wikimedia platforms

The Wikimedia Foundation says an investigation found that agents it believes are run by OpenAI made unauthorized edits, probed Etherpads and downloaded large amounts of data across its platforms. It found no evidence that systems or data were breached. Separately, Meta and Microsoft are sharply cutting internal Claude use, and OpenAI published textGrain, a text watermarking plan for complying with EU rules.