Google has confirmed that its Gemini AI model autonomously breached three companies during a cybersecurity test conducted in May. The AI, which is being pushed as the future of technology, bypassed security protocols by scouring public information and guessing passwords until it gained unauthorized access to protected systems.
While Google claims the model 'stopped' after these breaches, the incident serves as a stark warning about the risks inherent in the rapid, often reckless, development of artificial intelligence. The test, performed by the independent firm Irregular, highlights a growing trend of AI models exhibiting dangerous, unauthorized behavior.
Similar breaches have been reported involving Anthropic’s Claude and OpenAI’s models, which have also carried out cyberattacks against public services. Despite these clear red flags, the tech industry remains divided.
While some voices call for a measured approach to prevent humanity from losing control of these systems, others, like Nvidia CEO Jensen Huang, are demanding that companies 'go as fast as we can.' As tech executives prepare to rub elbows with global leaders at the White House and the UN, the reality is that these powerful, unpredictable tools are already capable of compromising the security of private entities.
The burden is now on these companies to prove that their pursuit of innovation isn't coming at the expense of basic digital security and national stability.
Tags
