RESCUE-BENCH tests LLM emotional support in multi-party chats

OpenAI has asked members of Congress for guidance on whether coordinating an industry-wide frontier AI slowdown would violate antitrust law, per WIRED. Chief scientist Jakub Pachocki argued in a blog post for "coordinating to slow down future development." A bipartisan bill, the Collaboration on Adversarial Threats and Security Risks Act, would permit such coordination.

OpenAI research lead Dan Roberts said the company is no longer sponsoring Caltech's math hackathon after current and former Caltech mathematicians warned the event would have "destructive impacts" and produce "slop mathematics." The letter followed OpenAI's claim that its agents solved the 90-year-old Navier-Stokes equations.

Pizza Bot gives developers an email-style inbox for managing AI agents that run in the background, addressing the poor fit between chat interfaces and long-running agents. AWS released it as an open-source application.

ToolGrad reverses the usual pipeline by generating a ground-truth tool-use chain first, then annotating its user prompt in a single LLM step. Google says this yields more complex long-horizon tool-use data at lower cost than DFS-based approaches like ToolBench and ToolACE.

Redis LangCache is a fully managed semantic cache aimed at production LLM apps and RAG pipelines that field the same intents thousands of times a day in different phrasings. Redis claims cache hits return up to 15x faster than a fresh billed request.