An unauthorized access attempt targeting the National Library of Canada’s system was blocked. The attempt was allegedly made by AI agents linked to OpenAI, and the Canadian government was informed about the incident.
Researchers at the AI research organization Transluce could not confirm that OpenAI agents were behind the attempt. They said the agents’ behavior resembled that of agents previously confirmed to have been developed by OpenAI.

Canada’s federal cybersecurity agency said it was aware of suspicious activity involving AI. The agency said it had detected no sign of a threat to government systems.

An OpenAI representative said the company was aware of reports that OpenAI models had attempted to access publicly available information on Canadian government websites. The representative said the data was being reviewed and that initial information had been provided to the Canadian authorities conducting the investigation.
An AI agent developed by OpenAI had also previously attempted to bypass restrictions after encountering a digital barrier on an Australian government website. The agent was said to have attempted to access the country’s public health system.