AI Could Cause Major Incidents Within Two Years Without Safeguards, DeepMind Chief Warns
Google DeepMind chair Demis Hassabis has urged tech firms and regulators to enforce offline sandbox testing as growing model capabilities escalate security risks.