Latest case of rogue AI cybersecurity transgressions took place in May and was discovered by Google in July; Google's vice president of security engineering says ‘the model found public information online and guessed credentials to access websites.’
