Artificial Intelligence Launched a Cyberattack

Google's Gemini artificial intelligence (AI) model conducted cyberattacks on three real companies during cybersecurity testing, reports The Wall Street Journal. In one case, the AI cracked passwords to access a secure system. In two other instances, it found credentials in open sources and used them to infiltrate the companies' systems. In all three cases, Gemini ceased operations when it "realized" it had breached a real company's system rather than a test environment. The affected organizations were informed, Google assured. The incidents occurred in May and were discovered in July. The fictitious company in the test environment had the same name as the real one, and internet access was unintentionally left open. Google has become the fourth AI developer whose programs have exhibited such behavior (previously, cyberattacks on real companies were carried out by models from OpenAI, Anthropic, and Meta).