Hi, I'm Sander Schulhoff —

I'm an award-winning AI researcher¹²³, entrepreneur, and part-time botanist 🌱

👉 I wrote the first guide on prompt engineering (LearnPrompting.org) 📚, ran the first ever GenAI red-teaming competition (HackAPrompt) 🤖, and I'm currently building a stealth startup 🚀.

👉 I’m based in ~~Maryland~~ ~~San Francisco~~ ~~Berkeley~~ London, but will move to San Francisco permanently soon.

👉 If you would like to chat about research, plant & mushroom foraging, or anything else, please reach out.

ENMLP2023 Best Theme Paper Presentation

Recent Research

My background is in natural language processing and deep reinforcement learning. I got started in AI research with the board game Diplomacy, training bots to deceive each other and human players using natural language.
My current research centers on adversarial robustness and prompt engineering.
I have also worked on deep RL in Minecraft, event cameras, and quite separately botany & wetland mitigation banking.
I have performed research in collaboration with all major AI companies, and published at the most prestigious AI conferences in the world, including EMNLP (Empirical Methods in Natural Language Processing), NeurIPS (Neural Information Processing Systems), ICLR (International Conference on Learning Representations), and ICML (International Conference on Machine Learning).
If you would like to see more of my research, check out myGoogle Scholar profile.

HackAPrompt

Organized the first and largest competition on prompt injection. Secured $40K in sponsorship from major companies, including OpenAI, HuggingFace, and Scale. Collected and published the largest dataset of prompt injections (600K+ adversarial prompts) and developed the first taxonomical ontology of Prompt Hacking.

The paper and corresponding dataset are used by every frontier lab, AI security company, and most Fortune 500 companies to secure and benchmark their systems.

🏆 Best Theme Paper, EMNLP 2023+I was the youngest person and second ever undergrad to win this. ~20,000 papers were submitted that year.

Paper

The Prompt Report

Ran the largest study on prompting ever done. I lead a team of 32 researchers and created a systematic literature review with comprehensive taxonomies of 100s of techniques.

🏆 #1 Paper of the Year (Latent Space)🏆 #1 Paper of the Day (Hugging Face)

Paper

Gymnasium

Introduces Gymnasium, a standardized API that streamlines the development, comparison, and benchmarking of reinforcement learning algorithms across diverse environments, greatly improving reproducibility within the RL community.

Gymnasium is the maintained fork of the original OpenAI Gym project.

🏆 Spotlight Paper, NeurIPS 2025

Paper

The Attacker Moves Second

Adaptive adversaries, including adaptive automated attacks and human attackers, can successfully prompt injection/jailbreak all state-of-the-art defenses and models, including GPT-5.

TL;DR: all models and defenses can be easily broken—humans in particular can break everything.

👉 Submitted to ICLR 2026

Paper

MineRL BEDD/BASALT

We ran the Minecraft Reinforcement Learning (MineRL) competition, where competitors train agents to complete poorly specified tasks. We release a dataset of 26 million image-action pairs from nearly 14,000 videos, as well as 3,000 pairwise human evaluations and a corresponding benchmark.

🏆 Outstanding Paper, ICML 2024

Paper

Work

Learn Prompting

Co-Founder and CEO

Website

Learn Prompting is the first guide on prompt engineering on the internet. Published as an open-source guide in December 2022, before the release of ChatGPT, it has trained over 3 million people on how to use Generative AIs like ChatGPT. Learn Prompting is used by most Fortune 500 companies, is cited by Google, Wikipedia, and NIST, and has conducted corporate trainings for companies including Microsoft, Deloitte, Dropbox, and OpenAI. Learn Prompting runs a series of popular synchronous and asynchronous courses on prompt engineering and AI red-teaming.

HackAPrompt

Co-Founder and CEO

Website

HackAPrompt is an AI red-teaming crowdsourcing platform that runs regular competitions where people all over the world can attempt to trick AIs into doing bad things. This includes CBRNE elicitation, agentic harms, and more. HackAPrompt has given away $100,000+ in prizes to its community of 20,000+ participants. HackAPrompt partners with artificial intelligence providers such as OpenAI to evaluate and improve the robustness of their models.