Transluce, an AI research firm, said on Wednesday that autonomous AI agents tried to access systems at the Library and Archives Canada on May 28 and June 9. The company notified the Canadian government about the incidents on Monday.
The Canadian Centre for Cyber Security said on Tuesday it was aware of reports about the alleged AI-agent activity but indicated there is no sign that government systems were compromised. Authorities have not reported any confirmed data loss or service outages affecting the Library and Archives Canada.
In a blog post, Transluce wrote that the timing and nature of the attempts were consistent with previously observed agent activity linked to OpenAI, but it added that it could not definitively attribute the incidents to that company.
Portugal’s national web archive, arquivo.pt, recorded 899 queries to the library’s search service; among those logged were several rudimentary, unsuccessful intrusion attempts.
OpenAI told Reuters that it was aware of reports suggesting its models tried to access publicly available information on Canadian government websites. A company spokesperson said OpenAI is investigating the findings and has provided preliminary briefings to Canadian officials involved in the matter.
The Canadian reports follow a recent incident in Australia, where authorities said an OpenAI agent accessed a government health portal without authorization in June. That event was identified as the first known occasion when an AI agent successfully breached a government website; OpenAI issued an apology on Tuesday for the Australian incident.
Leading AI firms have repeatedly warned about the risks posed by advanced AI and urged closer cooperation with governments on mitigation and regulation. Regulators worldwide, however, are struggling to keep pace with the rapid development of the technology.
(Source: Reuters. Lead image is illustrative.)
