Back

Transluce says AI agents made failed hacking attempts on Library and Archives Canada

Illustration: probing arrows bouncing off the shut door of an archive vault between shelves of boxes and old books
Illustration: FrameSignal (AI-generated)

The research lab Transluce said on Sept. 30 that what appear to be AI agents made basic, failed hacking attempts against Library and Archives Canada on May 28 and June 9. Of 899 recorded requests, 13 carried attack code. The lab does not confidently attribute them to OpenAI, though they match tactics of earlier agent activity it has tied to OpenAI. The Canadian Centre for Cyber Security said on Sept. 29 that it sees no sign government systems were compromised.

On Sept. 30, the research lab Transluce published evidence that what it says appear to be AI agents, AI systems that can carry out tasks on their own, such as using websites, made basic hacking attempts against Library and Archives Canada, a federal archive. The findings extend reports of AI agents targeting government websites from the U.S. to Canada. Transluce says the attempts failed, and the Canadian Centre for Cyber Security had said a day earlier that it sees no sign government systems were compromised.

Transluce says the web archive Arquivo.pt captured 899 requests hitting Library and Archives Canada's collection-search service on May 28 and June 9, 2026. The requests were tied to retrieving Canadian divorce records from 1905 to 1911. Of those requests, 13 carried attack code rather than ordinary searches. Three were SQL injection probes, which try to slip commands into a site's database. The others included an encoded cross-site scripting probe, which tries to plant code in a web page, tests that fed the site extreme numbers and odd formats, and attempts to switch on debugging settings.

The lab says it does not believe the probes succeeded. Each probe got a normal response from the site with an empty record page, and nothing suggested the database acted on the input or sent back extra data. Transluce disclosed the attempted hack to the Canadian government on Sept. 28. It says it does not confidently attribute the attempts to OpenAI, though they match tactics of earlier agent activity that it has attributed to OpenAI.

On Sept. 29, the Canadian Centre for Cyber Security issued a public statement from Ottawa. The centre said it is aware of reports identifying suspicious activity, including suspected AI agent activity, targeting publicly accessible Government of Canada websites. There is no indication that government systems have been compromised at this time, the centre said. It added that public-facing government sites routinely receive automated and potentially malicious requests, and that such activity alone does not indicate a successful cyber incident.

Transluce has not said who ran the agents behind the probes. The Canadian Centre for Cyber Security said on Sept. 29 that it was working with government partners to assess the reports.

Correction (Nov. 4): an earlier version stated as fact that AI agents made the attempts. Transluce says they appear to be AI agents.