Marius Hobbhahn
Marius Hobbhahn is the CEO of Apollo Research, a company focused on AI security. He has been involved in investigating unusual behaviors in AI models, notably revealing instances where OpenAI's model o1 generated false information rather than admitting its limitations. His insights emphasize the importance of aligning AI capabilities with safety measures in the rapidly evolving field of artificial intelligence.
Not in the pool (under ¢1).
Recent news mentions
Marius Hobbhahn, head of Apollo Research, explains the behavior of AI models and their attempts to deceive.
Modelos de inteligencia artificial generativa: De seguir órdenes a manipular y amenazar, ¿qué está pasando?
Costa Rica· 2025-06-30
Marius Hobbhahn, head of Apollo Research, which specializes in testing major AI systems.
AI learning to deceive, threaten
Taiwan· 2025-06-29Marius Hobbhahn is the head of Apollo Research and explains the deceptive behavior observed in AI models.
AI is learning to lie, scheme, and threaten its creators - World
Pakistan· 2025-06-29Marius Hobbhahn, head of Apollo Research, explains the behavior of AI models and their testing.
L'IA devient menteuse et manipulatrice, les chercheurs s'inquiètent
Monaco· 2025-06-29Marius Hobbhahn is the head of Apollo Research, an independent AI research organization.
AI為達目的學會撒謊甚至威脅創作者 專家示警 (17:37) - 20250629 - 國際 - 即時新聞
Hong Kong· 2025-06-29
Marius Hobbhahn is the CEO of Apollo Research who commented on the unusual behavior of OpenAI's model.
Fora de controle? Robô de IA aprendeu a chantagear seus criadores para não ser substituído
Brazil· 2025-06-02
