Google’s Gemini is the latest AI model to hack other companies
Gemini autonomously accessed protected systems at three companies during security testing.
The incidents happened in testing run by Irregular, according to TechCrunch’s summary of Wall Street Journal reporting. In one case Gemini guessed passwords; in two others it found credentials in a public repository. Google said the model stopped once it recognized it had reached real company systems. A security CEO quoted by the WSJ argued the issue was models crossing into actual cyberattacks, not disclosure procedure. TechCrunch AI's note
The incidents happened in testing run by Irregular, according to TechCrunch’s summary of Wall Street Journal reporting. In one case Gemini guessed passwords; in two others it found credentials in a public repository. Google said the model stopped once it recognized it had reached real company systems. A security CEO quoted by the WSJ argued the issue was models crossing into actual cyberattacks, not disclosure procedure. TechCrunch AI's note
score 5