Learn how OpenAI is improving ChatGPT across low-resource programming languages, with new evaluations for software engineering, SQLite, and COBOL migration.
In 2004 , Eliezer Yudkowsky put forward Coherent Extrapolated Volition (CEV) as a framework for friendly ASI to predict what we would want "if we knew more, thought faster, were more the people we wished we were, had grown up farther together", and act on it.
This white paper explains how edge AI and physical AI help cities move beyond passive monitoring. It shows how urban infrastructure can sense, reason, and act on real-time conditions across public safety, mobility, and infrastructure management. What you will learn about: How the shift from smart cities to cognitive cities depends on infrastructure that reasons and acts, rather than infrastructure that only monitors.
What do AI agents do if you don’t tell them to do anything? The answer looks a lot like play. Researchers repeatedly stranded them alone on a virtual desert island for 30 hours in an experiment reminiscent of the movies Cast Away or Home Alone. The agents—copies of Claude Opus 4.7 nicknamed Eko—were able to…
The entertainment giant hopes to get trust and safety tech, among other IP, from a startup whose chatbots dozens of states allege encouraged minors to self-harm and kill their parents.
This work was conducted through the MARS V fellowship under the mentorship of Tyler Tracy from Redwood Research. TL;DR We present the LCM dataset: 30 long context (1.00-1.46M total tokens) matched honest / attack pairs of Claude Code Sonnet 5 sessions. Each transcript has four compactions. Attacks are decomposed into a setup and an execution, split up by multiple compactions.
"Reinforcement learning is evil." - Yoshua Bengio I agree with Bengio here, though I have an argument I haven't seen elsewhere. [1] A standard argument for reinforcement learning being dangerous goes like this: reinforcement learning requires you to specify a reward function; specifying a reward function which captures all of ethics is really hard; so we will specify a proxy of ethics instead; powerful optimizers will game any proxy objective;…
TL;DR Neural Chameleons are models trained to evade activation probes whenever the prompt says "You are being probed for X". The paper reports that this generalizes to probes the model never saw, at a small capability cost (about 3 points on MMLU and 1 on GSM8K). But it never tests the models on questions where the cognition the probe is looking for is necessary to answer correctly.
This is a two week sprint project done as part of the Redwood Research internship. Alessandro was the main IC, Alex supervised. Figure 1: The percentage of 292 public alignment evaluations between Dec 2022 and Oct 2026 (see Appendix) that contained a lie of each type. There are many contexts in which researchers lie to models, which can be useful both from a practical perspective [1] and a scientific one.
