OpenAI pulled the plug on a cluster of accounts it says belong to a China-linked hacking group using ChatGPT to sharpen its attacks.
The company banned accounts tied to an operation it tracks as SweetSpecter, a suspected China-based threat actor. According to OpenAI, the group used its AI tools to research security vulnerabilities, write code, and support spear-phishing campaigns. OpenAI did not name specific targets or detail how many accounts were involved. The bans are part of the company's ongoing effort to curb malicious use of its models.
AI chatbots are becoming a force multiplier for attackers who once needed specialized skills to write exploit code or craft convincing phishing lures. That lowers the bar for espionage-style operations and puts AI vendors in the position of playing whack-a-mole with operators who can simply create a new account. It also puts the same tools sold to developers and businesses squarely inside the threat-intel conversation, not just the productivity one.
Banning accounts is a start, not a fix - the next SweetSpecter account is one signup away.