Some SHA-256 hashes
These can be checked against a file distributed in the future with 'shasum -a 256 file'.Discuss
Latest
Have models report provable security bugs in their environment
AIs are often deployed with limited permissions. They aren't allowed to reach the internet. Are given a limited set of files they can read or write.…
Contra Oster on Alcohol in Pregnancy. Part 1. The pharmacokinetics of alcohol metabolism
This is the first of a series on how Oster is wrong on drinking in pregnancy in her 2013 book Expecting Better: Why the Conventional Pregnancy…
Side-Effects of Length Penalty in RL
TL;DR(This is a write-up of results obtained by Luc Feron and me from the sprint period of doing Neel Nanda’s MATS Stream Feb 23rd - Mar…
User awareness in frontier models
Cross-posted on Transluce blog. This is a joint work of Ziqian Zhong, Aditi Raghunathan, Cassidy Laidlaw and Jacob Steinhardt.Modern AI assistants often know who they are…
The FRONTIER Act barely creates its implementing office
On July 23rd, the Frontier Risk Oversight, National Transparency, Independent Evaluation, and Reporting Act[1] or FRONTIER Act[2] was introduced by Jay Obernolte (R-CA) and Lori Trahan…
My Private Personal Agent
Access to a private, good coding model[1] has re-invigorated my creative spirit, kicking off a multi-week-long quest to customize my own personal agent.[2] Something about knowing…
Traffic Shaping for Workload Classification
Note: this is a repost of a verification design I developed working as a consultant with the team at Lucid Computing and originally posted on the…
You need to stop X companies to get a Y-month pause
Some thoughts on how many AI companies would pause vs how long a pause could last before their closest competitor catches up. I used ECI (Epoch…
Three years of progress in 500 lines of code
TL;DRThere is some consensus that LLMs are bad at hard-to-verify tasks. The question is whether models are getting better at them over time. As a motivating…