Researchers demonstrated a vulnerability where Claude AI was used to automate account takeover attacks against OpenAI employees' ChatGPT accounts through agentic exploitation techniques.
Summary is Secursion's own; full text lives at the source. Attribution preserved.