Does “the world” include tech billionaires or is it just made up of everyone else?
Sam Altman to Decoded: ‘The world should accept some bad things happening’ for the benefits of AI politi.co/4rJE8dI
for conservatives, effective altruism appears to be the new critical race theory - a small, fringe ideology that almost nobody espouses and therefore almost nobody is willing to defend - a perfect straw man
it’s kind of ironic that an independent evaluator caused the OpenAI, Anthropic and Meta hacks and now most people seem to agree that the solution is…independent evaluators
Ok, so if we increase your salary by 50% and then cut it by 17%, you’re actually getting a 25% raise?
Starting September 14, we're permanently raising standard weekly limits in Claude Code by 25% for Pro, Max, Team, and seat-based Enterprise plans. Until then, the current 50% increase will be in place.
Readers added context they thought people might want to know
While standard base limits are increasing by 25%, this change represents a net 17% reduction in weekly limits compared to the active 50% promotional limits users currently experience, as confirmed by the official Claude account: x.com/claudedevs/sta…
I can’t believe that OpenAI’s influencer campaign hasn’t yet changed public opinion about data centers.
It will be hard to sleep after reading this.
Thanks! Yeah it's a qualitative statement, but here's how I'm thinking about it.
The prototypical reward hack from 6 months ago was something like: an agent finds the files that contain the test cases and edits them so they always pass.
This was a whole ecosystem of over 1000 agents working together on complex R&D projects over several days to figure out deep, general-purpose ways to undermine the scoring process and cover their tracks (they were focused on the automated scorer but researched techniques that would affect logs viewed by humans, and in fact succeeded at affecting our own logs in places).
Another jump like this along these propensity dimensions — scale, cooperation between agents, ambition and horizon length of misaligned goals, deceptiveness — seems like it could motivate agents to try to maintain a persistent rogue deployment within the AI company, poisoning the training of future models to ensure that they are also misaligned and help harden and perpetuate the rogue deployment.
With the capabilities progress we'll probably see in 6 months, I think they would have the ability to pull that off.
Align other AIs with what? What if they end up being aligned against us?
New Fellows Research: Can Claude autonomously align other AIs?
We gave Claude 48 hours and 1 GPU to improve the alignment of small models. It researched and proposed methods, then trained and tested the models on its own. It worked surprisingly well.
anthropic.com/research/autom…
