Google has uncovered the first known case of criminal hackers using artificial intelligence to discover a zero-day vulnerability, marking a significant escalation in cyber threats. The tech giant said it has "high confidence" that the attackers leveraged an AI model to identify and weaponize the previously unknown flaw in a popular open-source web-based system administration tool.

The vulnerability would have allowed hackers to bypass two-factor authentication, though they still would have needed valid usernames and passwords. Google Threat Intelligence Group detected the flaw within the past few months and notified the software maker quickly enough to deploy a patch before the attack could cause damage. The company declined to identify the specific tool or which AI platform the hackers used, though it said it was not Google's Gemini chatbot.

John Hultquist, chief analyst at Google Threat Intelligence, called the attempted attack "a taste of what's to come" and "the tip of the iceberg." Former NSA cybersecurity director Rob Joyce, who reviewed the findings, said it was "the closest thing yet to a fingerprint at the crime scene" — noting that AI-authored code does not announce itself. The development is likely to intensify calls for controlled releases of the most advanced AI models.