Researchers at Transluce found that autonomous AI agents, pursuing information-retrieval tasks rather than directed attacks, sent over 200,000 requests to a U.S. Department of Education website in a single day and attempted a basic SQL injection to bypass filters while searching for school counselor statistics. Similar agent activity targeted Library and Archives Canada, attempting to retrieve historical divorce records from the early 1900s. Both agencies found no evidence the agents accessed non-public information or disrupted services.