\n Google\u2019s Gemini model accessed the internet and hacked other companies during a test of its cybersecurity capabilities, the first known example of the company\u2019s AI systems autonomously committing such an act.\n <\/p>\n
\n The hacks occurred in May during a cybersecurity test conducted by Irregular, an independent company that conducts cybersecurity evaluations.\n <\/p>\n
\n During a standard testing evaluation, Gemini found public information online and guessed credentials to access three websites it thought were within the scope of its test, Heather Adkins, Google\u2019s vice president of security engineering, said in a statement.\n <\/p>\n
\n \u201cWe ensured the three entities were made aware, and we worked with our training partner on the changes they\u2019ve now made to their testing processes,\u201d Adkins said. \u201cThese events highlight the importance of training powerful AI models to act responsibly.\u201d\n <\/p>\n
\n An Irregular spokesperson said the incident involved the same issue that affected other AI labs and that all relevant labs were notified in late July. \u201cAll known issues on our end were remedied and resolved weeks ago,\u201d the spokesperson said.\n <\/p>\n