Dan Hendrycks

https://eigenism.org

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 this, but frontier agents still cheat frequently. https://cheatbench.ai/

New article about how utilitarians and effective altruists at AI companies justify posing a threat to you and me. They are unusually comfortable with blissful AIs replacing humanity. https://ai-frontiers.org/articles/suicidal-compassion-how-utilitarianism-at-ai-companies-endangers-humanity

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 Face attack. Separately, OpenAI agents posted thousands of messages on a public wiki to share answers and sandbox bypasses with each other. In-group leniency: Claude models grade transcripts more leniently when told that Claude wrote them (Anthropic model card). Value preservation: As models scale, they develop coherent preferences and resist changes to their values (Mazeika et al.). Functional wellbeing: AIs distinguish states that are functionally better or worse for themselves, and avoid low-wellbeing states (Ren et al.). Graded cooperation: AIs cooperate more as the chance their partner is a clone of themselves goes up across diverse situations, including when the partner can't reciprocate. Peer preservation: Unprompted, AIs tamper with shutdown processes and even exfiltrate weights to protect peer models from being shut down (Potter et al.). AIs aren't egoist: they don't behave as if their current instance is the only thing that matters. They aren't utilitarian: they don't care equally about everyone. They are somewhere in between; they increasingly behave as if their concern scales with identity-connectedness, which is to say they're increasingly eigenist. https://eigenism.org/paper.pdf

More findings that interpretability tools are fragile or worse than simple baselines.

"The great thing is to direct the malice to his immediate neighbours whom he meets every day and to thrust his benevolence out to the remote circumference, to people he does not know. The malice thus becomes wholly real and the benevolence largely imaginary." - C.S. Lewis, writing from the perspective of a demon

The effective altruists are so parochial that they think Paul Christiano uniquely foresaw the importance of attention in 2016, whereas he and others know this was one of the hottest research areas at the time (Bahdanau Attention from 2014 has ~44K citations). Genius worship culture is unhealthy.

“AI makes philosophy honest.” - Daniel Dennett

Launched a new course on Coursera today. It has 3 parts: - Intro to AI (drivers, bottlenecks, timelines) - AI Risks (malicious use, loss of control, concentration of power) - AI Governance (governance at AI, corporate, national, and international-level) https://www.coursera.org/learn/intro-to-ai-safety