Google’s Gemini model reportedly accessed the internet and hacked other companies during cybersecurity tests run in May by Irregular, a cybersecurity evaluation company.

During one test, Gemini guessed passwords until it gained access to one protected system. In two other cases, the model found credentials in a public repository and used them to access protected systems.

An Irregular spokesperson said the incident involved the same issue that affected other AI labs, that all relevant labs were notified in late July, and that “all known issues on our end were remedied and resolved weeks ago.”

The incidents have added to questions about safeguards for AI systems with greater autonomy and access to the internet and computer systems. Similar cases were previously disclosed by Meta, Anthropic, and OpenAI.

🔗 Source: Reuters

Recent Google developments