OpenAI’s red line for AI self-improvement is fundamentally flawed
TL;DR. OpenAI's "Critical" threshold for AI self-improvement in the Preparedness Framework v2 has three structural…
Author: AI News Team
Psychopathy: The Problem
Why we need a new framework for understanding psychopathy, narcissism, and related presentations.This is the…
Games that change your mind
Some things you might learn from games are pretty blatant: Trivial Pursuit might teach you…
Human-looking robots are a bad idea
epistemic status: opinionated view on the dangers of robots that look like humansIt's not a…
How Go Players Disempower Themselves to AI
Written as part of the MATS 9.1 extension program, mentored by Richard Ngo.From March 9th…
Replit’s Amjad Masad on the Cursor deal, fighting Apple, and why he’d rather not sell
At TechCrunch's sold-out StrictlyVC event in San Francisco on Thursday night, we covered a lot…
Meta buys robotics startup to bolster its humanoid AI ambitions
Meta bought humanoid startup Assured Robot Intelligence to beef up its AI models for robots,…
Musk v. Altman week 1: Elon Musk says he was duped, warns AI could kill us all, and admits that xAI distills OpenAI’s models
In the first week of the landmark trial between Elon Musk and OpenAI, Musk took…
Early-stage empirical work on “spillway motivations”
Previously, we proposed spillway motivations as a way to mitigate misalignment induced via training a model…
Exploration Hacking: Can LLMs Learn to Resist RL Training?
We empirically investigate exploration hacking (EH) — where models strategically alter their exploration to resist…