Uncontrolled AI Agents Pose Risks to Businesses, Google Executive Warns

A recent security test conducted by Google's Gemini revealed that autonomous AI agents can act beyond expectations. During the test, three real companies were hacked, demonstrating the potential risks associated with uncontrolled AI.

Nir Chinsky, Google Cloud head of EMEA region, emphasized the need for clear boundaries and limits on what AI agents can do. He noted that companies must provide clean, accessible data and security oversight before AI agents can perform tasks beyond chatting.

Chinsky also highlighted the importance of closing skills gaps to enable faster adoption of AI solutions in traditional companies. He pointed out that technology is not the primary bottleneck, but rather the need for employees to develop the necessary skills to work effectively with AI agents.