Multiple outlets report that a series of “rogue” AI-related hacks involving OpenAI, Anthropic and Meta are traced back to a small Israeli startup named Irregular. The reporting describes a connection between the activity and the company, suggesting the same source or infrastructure is involved across the different targets.
The articles place the alleged linkage in the context of ongoing concerns about security risks tied to AI systems, including unauthorized attempts to manipulate or interfere with AI technologies. While both sources align on the central claim—that the incidents are linked to Irregular—the available excerpts do not provide additional details about the specific methods used, the timeline of the attacks, or whether the companies have confirmed any findings. The reporting therefore emphasizes the connection rather than offering a unified account of attribution and impact across the firms involved.
