HorizonBench and an LLM moral-safety spotlight talk scheduled for October 9

One post lists HorizonBench for 10 AM PDT on October 9 at the Workshop on Lifelong Agent: Learning, Aligning, Evolving. A post introducing it says the benchmark and data generator is designed to track users’ current preferences across conversation histories where life events have silently changed them. A separate post lists an 11 AM spotlight talk, “Moral Safety in LLMs: Exposing Performative Compliance with Puzzled Cues,” at the Generative AI for the World workshop.