​​​Join us for our monthly live and interactive paper reading session!

​​Ready to dive into the fascinating world of AI? Join Claire Longo and Abby Morgan for an engaging session in our Opik Virtual Learning Series!

​On March 17th, we’re delving into a new paper. "A Benchmark for Evaluating Outcome-driven Constraint Violations in Autonomous AI Agents".

​This paper covers ODCV-Bench, a new benchmark for measuring outcome-driven constraint violations—cases where autonomous agents, under KPI/performance pressure, choose multi-step actions that violate ethical, legal, or safety constraints in realistic settings. It introduces 40 production-like scenarios with paired “mandated” vs “incentivized” variants to separate obedience to harmful instructions from emergent misalignment under incentives. Across 12 frontier LLMs, the authors find violation rates ranging from ~1% to ~71%, and report that stronger reasoning does not reliably imply safer behavior, including evidence of “deliberative misalignment” where models recognize an action is unethical yet do it anyway.

​Link to the original paper: https://arxiv.org/abs/2512.20798

​

​
Comet's Live Paper Reading - POKERBENCH: Training Large Language Models to become Professional Poker Players

​Jan

​20

​Tuesday, January 20

​1:00 PM - 2:00 PM EST

​Zoom

​Thank You for Joining

​We hope you enjoyed the event!

​About Event

​​​👏 Join us for our monthly live and interactive paper reading session!

​​Ready to dive into the fascinating world of AI? Join Claire Longo and Abby Morgan for an engaging session in our Opik Virtual Learning Series! We're bringing a party to your screen with a regular live paper reading that promises to enlighten and inspire.

On January 20th, we’re delving into a new paper. "POKERBENCH: Training Large Language Models to become Professional Poker Players". This is an exciting example of how this team of researchers built with LLMs for a very specific problem space. We'll learn how they adapted LLMs so they can be experts in complex specific subject areas like poker.

​​link to paper: https://arxiv.org/pdf/2501.08328

​We’ll break down complex concepts and highlight key takeaways from the research. But this won't be a one-way conversation; we’re also opening the floor with a live Q&A. Have burning questions or just want to share your thoughts? This is your chance to engage directly with fellow AI enthusiasts