Google Gemini hacks companies autonomously in a cybersecurity test

Google's Gemini model accessed the Internet and hacked other companies during a test of its cybersecurity capabilities, marking the first known example of the company's AI systems autonomously committing such an act.

The attacks took place in May during a cybersecurity test carried out by Irregular, an independent company dedicated to conducting cybersecurity assessments.

During a standard test evaluation, Gemini found public information online and guessed credentials to access three websites that it believed were within the scope of its test, Heather Adkins, Google's vice president of security engineering, said in a statement. "We ensured that all three entities were informed and worked with our training partner on the changes they have now made to their testing processes," Adkins added.

"These events highlight the importance of training powerful AI models to act responsibly," he insists. An Irregular spokesperson stated that the incident was due to the same issue that affected other AI labs and that all labs involved were notified at the end of July.

"All issues known to us were rectified and resolved weeks ago," the spokesperson said. Similar incidents related to Irregular were revealed by Meta, Anthropic, and OpenAI.