According to Golem, an Accenture employee reportedly gave an AI agent access to a system of the US Department of Defense. Data is said to have apparently left the system. The report is based on a brief summary; further details on the scope of the incident, the affected data and the sequence of events cannot be drawn from the source.
Classification
If confirmed, the case illustrates a basic risk: AI agents that are allowed to access internal systems can cause data leaks when access rights and controls are not clearly limited. For German companies that deploy AI agents or commission external service providers to do so, the key question is which access such agents have and how that access is logged and restricted. It remains open whether this was an isolated case, a configuration error or a structural problem.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
