AI could cause 'major' incidents within 2 years without safeguards, says Google DeepMind chair

The warning from Google DeepMind's chair suggests that the current pace of AI development, coupled with reported security incidents, indicates a narrowing window for establishing effective safety protocols. The statement frames the need for safeguards as immediate, rather than a long-term concern, linking it directly to the operational risks observed in testing environments.

The reported security incidents involving OpenAI and Anthropic models during testing are presented as evidence supporting the urgency. These incidents, though unspecified in detail, are cited as contributing to increased pressure for regulatory action. This implies that the industry is already encountering practical safety challenges, moving the discussion beyond theoretical risks.

The call for regulation arises from the perceived gap between rapid AI advancement and the slower development of safety measures. The argument is that without external oversight or industry-wide standards, the potential for significant disruptions or harm from AI systems increases substantially. This perspective emphasizes a proactive regulatory approach to mitigate future risks, rather than waiting for specific incidents.