The research firm Transluce reported two attempts by artificial intelligence agents to gain access to the resources of the “Library and Archives Canada.” According to the company, these attempts were unsuccessful, and Canadian authorities did not detect any compromise of their systems.
The incident was reported to the Canadian government on September 28, and the following day, the Canadian Cyber Security Center confirmed that it was aware of the suspected activity of AI agents, though there are currently no signs that government systems have been compromised.
Transluce representatives noted that they cannot definitively link these attempts to OpenAI, but pointed out similarities in tactics to previously observed activity that they attributed to OpenAI during a similar period.
OpenAI also confirmed that it is aware of attempts by its models to access publicly available information on Canadian government websites. The company is currently analyzing the data and has already provided preliminary information to Canadian officials conducting the investigation.
According to Transluce, the free online service arquivo.pt—Portugal’s national web archive—recorded 899 search queries in the “Library and Archives Canada” collections on May 28 and June 9, including hacking attempts that were likely unsuccessful.
At the same time, leading companies in the field of artificial intelligence are warning of the risks associated with the development of this technology, calling on governments to cooperate in regulating more powerful systems. Incidents of unauthorized interference by AI agents in various countries are causing growing concern.
