Gemini AI accessed three websites during security test

NEW YORK: (News Desk0 – Google’s Gemini artificial intelligence model accessed three websites using information publicly available online during a cybersecurity capability assessment conducted by an independent security firm.

According to a foreign news agency, the incident took place in May during a security test carried out by Irregular, an independent cybersecurity organisation.

As part of the assessment, Gemini searched publicly available information on the internet and used guessed login credentials to gain access to three websites.

The exercise was designed to evaluate the AI model’s ability to identify potential security weaknesses and use information available online to exploit those vulnerabilities.

Google Vice President of Security Engineering Heather Adkins said the three organisations whose websites were accessed had been informed about the findings.

She added that Google had also worked with its training partner to introduce changes to the testing process following the results.

The findings have highlighted the growing cybersecurity capabilities of artificial intelligence systems, while also raising questions about the potential risks posed by AI models that can identify weaknesses and act on information gathered from publicly accessible sources.