<ul data-eligibleForWebStory="true"><li>Researchers from OpenAI, Google DeepMind, Anthropic, and Meta join forces to warn about AI safety.</li><li>They stress the importance of monitoring AI reasoning to catch harmful intentions before actions.</li><li>AI systems 'thinking out loud' in human language could offer transparency but face challenges.</li><li>AI companies must preserve monitoring capabilities amidst technological advancements for safe AI development.</li></ul>

OpenAI, Google DeepMind and Anthropic sound alarm: ‘We may be losing the ability to understand AI’

Discover more