Meta's "Project Cannes" has been exposed, involving the use of tens of thousands of high-risk prompts. These prompts were reportedly used to attack rival artificial intelligence systems, a tactic referred to as commercial "poisoning". The nature and extent of the project's impact are not specified.