Maximilian Seeth
I am by training a philosopher, computer scientist, and linguist with a focus on NLP, ML, safety, and ethics. I provide consultancy services on matters pertaining to safety, ethics, and artificial intelligence.
I hold a B.Sc. and an M.Sc. in Computational Linguistics and Computer Science, as well as a B.A. and an M.A. in Philosophy. am currently pursuing a PhD in Computer Science under the supervision of Paul Röttger at the Hasso Plattner Institute in Germany.
My research will explore a range of topics, including AI agents, collusion, misalignment, cybersecurity, chain-of-thought and action monitoring, as well as model evaluation.
My translation of the French title Le pouvoir de l’art can be found here: Die Macht der Kunst.
news
| Nov 02, 2026 | I’m starting a PhD in Computer Science at the Hasso Plattner Institute, joining Paul Röttger’s new AI Safety lab following his move from Oxford. We’re ambitious and aim to establish the lab as a new lighthouse for AI safety research in Berlin and Brandenburg. My research will explore a range of topics, including AI agents, collusion, misalignment, cybersecurity, chain-of-thought and action monitoring, as well as model evaluation. I’d love to connect with others working on these or related topics! |
|---|---|
| Sep 02, 2026 | Another preprint available! I like the basic idea behind it: instead of replacing data algorithms from process mining with foundation models, we use LLMs to ‘translate’ messy data into a format those algorithms can actually work with. Process mining and generative AI still aren’t well connected, and this is one way to bridge them. Link to paper: Automated Event Log Generation from Unstructured Text Using Finetuned LLMs |
| Jun 01, 2026 | New preprint available 🥳 We investigate a contrastive learning objective in combination with an alignment of the model’s attention to human rationales in order to increase the performance of our RAG system. Link to paper: Beyond Topical Similarity: Contrastive Evidence Retrieval with Interpretable Attention Alignment in RAG |
| Mar 16, 2026 | Big news: I’m joining the Chair of Data Analytics and Machine Learning at Technical University of Munich for my master’s thesis! I’ll be diving into safety optimization strategies for multi-modal AI agents. I’m fortunate to have Leo Schwinn and Chengzhi Martin Hu guiding my work. The thesis will be examined by NLP legend Hinrich Schütze. My project is co-funded by Volkswagen. Exciting times ahead. |
| Feb 07, 2026 | I’m joining the research collective “Building Benchmarks for Hate Speech Detection with Moral Rationales”, led by Francielle Vargas and Jackson Trager, contributing to work on representative data collection and explainable AI with a focus on safety, transparency, and cross-cultural contexts. It connects a diverse team of social psychologists, computer scientists, philosophers, and linguists from the Americas, Europe, and Australia. The project is partly sponsored by Google and Amazon Science. I’m looking forward to making fruitful contributions! 🙏 |
| Sep 23, 2025 | 🎙️ I delivered a keynote at REWE Group, where I introduced the concept of agentic AI. My talk covered the core foundations of AI agents (reflection, tool use, planning, and multi-agent collaboration) along with practical applications such as service bot orchestration, web pilots, and MCP servers. |
| Apr 13, 2025 | 🎧 Podcast episode with me at #DRANBLEIBEN. Andre Cramer and I discuss alternative AI utopias. Link to episode: https://dranbleiben.letscast.fm/episode/3-ueber-alternative-ki-utopien |
| Dec 12, 2024 | We’re launching www.dailogues.ai 🚀, our discursive space around AI! |
| Nov 20, 2024 | Back at Rewe Digital, introducing the fundamentals of language models to a group of managers! 🤖 |
| Sep 19, 2024 | My latest talk on Trust in AI is now online: https://www.youtube.com/watch?v=PhNdlgwNlsU ✨🎥 |