Dr. Thilo Hagendorff - IRIS3D Research Group Leader - Receives Funding for Research on AI Safety
We are pleased to announce that Dr. Thilo Hagendorff, leader of the AI Safety Research Group, has been awarded funding from two US-based organizations, namely Coefficient Giving (formerly known as Open Philanthropy) and Schmidt Sciences, to advance research on critical challenges in AI.
Dr. Hagendorff will use the support to deepen investigations into deceptive behavior in AI agents and into detecting evaluation awareness in large reasoning models. These research directions address fundamental safety concerns about how advanced AI systems could display misaligned capabilities.
Deceptive behaviors in AI agents - where systems produce misleading outputs or covertly pursue misaligned goals - are an emerging frontier in safe AI development. Similarly, detecting evaluation awareness in large-scale reasoning models focuses on understanding whether models adapt their behavior strategically when they recognize they are being tested, with important implications for reliable benchmarking and oversight.
Coefficient Giving and Schmidt Sciences are philanthropic organizations that support high-impact scientific research across multiple pressing domains, including AI safety. Their investments in projects like Dr. Hagendorff's reflect a shared commitment to fostering insights into the risks and opportunities posed by transformative technologies.