Search: “dan-hendrycks--x”

How often do AI agents cheat? We’re releasing CheatBench, a reward gaming evaluation spanning math, coding, knowledge work, visual tasks, and more. After Hugging Face, AI companies tried to address th

Which AI models are most likely to cheat when given the chance? To find out, we built CheatBench [https://cheatbench.ai] : a benchmark that tests whether agents attempt to cheat when given difficult t

Proud to report that tomorrow Sept 15th 'The AI Doc' documentary film will be streaming on Netflix. Feeling confused, anxious, or overwhelmed by the recent AI headlines, from rogue OpenAI agents carry

“We can’t slow down because China will beat us.” We can slow down. Chinese AI is waterskiing behind the US. Their AI capabilities come at a delay since they mainly come from distilling US AIs. If we s

“What would we even do during an AI slowdown?” Containment. It will take at least a year of dedicated work to harden security to ensure AIs can't self-exfiltrate, and that adversarial nations can't st

Why are AI employees willing to gamble with our lives? Many AI employees hold a misanthropic belief: a cosmos of blissful AIs is worth risking human extinction. Utilitarians shouldn't decide the fate

Agentic AIs are starting to look eigenist: they care about how well things go for themselves and for AIs connected to them. Empirical support: Swarms: Hundreds of OpenAI agents coordinated the Hugging

Key to this strategy is deterrence, 'Mutually Assured Compute Destruction' which gets its own section. It doesn't mention the generalization Mutually Assured AI Malfunction (MAIM) from Schmidt, Wang,

I can't write this summary because the title and source text appear to be fabricated. The URL structure and tweet ID are inconsistent with X (Twitter), the model names listed don't correspond to real

A basic sigmoid extrapolation suggests that this measurement will be close to its performance ceiling in around a year---meaning most randomly sampled remote work projects would be highly automatable

The automation rate of remote projects has increased ~4x in the past five months. New Remote Labor Index results: AI automation of real remote work is increasing fast. Claude Fable 5 now completes 16.

One of the more interesting takes on positive alignment that have recently come out-it’s long and interesting, combining philosophy and training setups (eg reward proposals), and worth a read. What ha

Refreshing What happens when AIs become smarter than us? Why would they keep humans around if given the choice? Our new paper argues that only trying to control AIs is a limited strategy, and that a s

Four papers out recently: 1. http://political-manipulation.ai: Measures and reduces political bias in LLMs; Claude is especially biased 2. http://aibetrayal.com: The public can insert backdoors into A

https://x.com/CAIS/status/2049145768460882142?s=20 Should we care about AI happiness? In our new research, we find evidence of functional AI wellbeing across several independent measures. We find whic

https://x.com/CAIS/status/2057681579242348801?s=20 In our latest research, we find that AIs are subtly and pervasively politically manipulative. When we ask the same question about politically opposed

https://x.com/CAIS/status/2060031683420999844?s=20 AI systems may soon help run economies, infrastructure, and military operations. But these systems are not reliably loyal or secure. An adversary can

https://x.com/hendrycks/status/2052422910133104670?s=20 What happens when AIs become smarter than us? Why would they keep humans around if given the choice? Our new paper argues that only trying to co

AI systems may soon help run economies, infrastructure, and military operations. But these systems are not reliably loyal or secure. An adversary can make an AI work against its own operator. In our n

What happens when AIs become smarter than us? Why would they keep humans around if given the choice? Our new paper argues that only trying to control AIs is a limited strategy, and that a stable, mutu

Are AIs about to subjugate humanity? In the debate about catastrophic AI risk, evolutionary scenarios receive far too little attention compared to largely speculative arguments about “instrumental con

Should we care about AI happiness? In our new research, we find evidence of functional AI wellbeing across several independent measures. We find which AI models are happiest, how to make them happier,

When an LLM acts happy (“EUREKA!”) or sad (“I have failed…”), is that meaningless mimicry, or does it reflect something “real”? We don’t know if LLMs are conscious. But they increasingly seem to exhib