Over the previous couple of months, Google Risk Intelligence Group (GTIG) has noticed menace actors utilizing AI to assemble data, create super-realistic phishing scams and develop malware. Whereas we haven’t noticed direct assaults on frontier fashions or generative AI merchandise from superior persistent menace (APT) actors, we have seen and mitigated frequent mannequin extraction assaults (a sort of company espionage) from personal sector entities all around the world — a menace different companies’ with AI fashions will seemingly face within the close to future.
At present we launched a report that particulars these observations and the way we’ve taken motion, together with by disabling related accounts as a way to disrupt malicious exercise. We’ve additionally strengthened each our safety controls and Gemini fashions towards misuse.
Learn the total report on the Google Cloud Threat Intelligence blog.
