2026-05-25T19:18:08Z https://yabu.me Raphaël Millière https://yabu.me/npub1a88y0h4ssc5lqk95x0l5sp5lunyhh9kfdj9aa6j3urmafftzw4vqhck4yy https://cdn.masto.host/sigmoidsocial/accounts/avatars/109/298/065/731/563/682/original/9dc51c7f64a33889.jpg https://cdn.masto.host/sigmoidsocial/accounts/avatars/109/298/065/731/563/682/original/9dc51c7f64a33889.jpg https://yabu.me/nevent1qqs9d3ecsprx2r95drh0gyjq959dccf80m4c7q6gzkv4azlc4s5agpqzyr5uu377kzrznuzcksel7jqxnljvj7uke9kghhh228s00499vf64szzm6jr A personal update -- I&#39;m happy to share that I&#39;ll be joining Oxford this fall as an associate professor, as well as a fellow of Jesus College and affiliate with the Institute for Ethics in AI. I&#39;ll also be establishing my AI2050 Fellowship from Schmidt Sciences there. Looking forward to getting started! 2025-08-21T14:09:02Z https://yabu.me/nevent1qqsq9p463nxdd74z4eh5vhr8n83d6g2z02yjve2xywwa4akedp3qqyszyr5uu377kzrznuzcksel7jqxnljvj7uke9kghhh228s00499vf64sdegcug Despite extensive safety training, LLMs remain vulnerable to “jailbreaking” through adversarial prompts. Why does this vulnerability persist? In a new open access paper published in Philosophical Studies, I argue this is because current alignment methods are fundamentally shallow. 🧵 1/13<br/><br/><a href="https://link.springer.com/article/10.1007/s11098-025-02347-3">https://link.springer.com/article/10.1007/s11098-025-02347-3</a><br/> <img src="https://cdn.masto.host/sigmoidsocial/media_attachments/files/114/659/354/097/169/978/original/8ca0c5cc201da6e8.png"> <br/> 2025-06-10T13:42:03Z