Learn to use AI

# AI safety

AI safety is a unit of 8 lessons in Rabbithole's Learn to use AI course, at levels 1 to 2.

## Lessons

1. AI safety: what it is and how people think about it · Level 1
2. Red-teaming: attacking an AI on purpose to find its flaws · Level 2
3. Jailbreaks: Getting an AI to Break Its Own Rules · Level 2
4. Toxicity Evaluation: judging harmful text · Level 2
5. Data Poisoning: Bad Data In, Bad Behavior Out · Level 2
6. Model Weight Security · Level 2
7. Testing AI for dangerous abilities · Level 2
8. Safety frameworks: how AI companies decide to launch · Level 2

## Sources

These lessons draw on 18 sources. The most used:

- [GPT-4 Technical Report](https://arxiv.org/abs/2303.08774)
- [Anthropic Risk Report, August 2026](https://anthropic.com/aug-2026-risk-report)
- [Frontier Safety Framework Report — Gemini 3 Pro](https://storage.googleapis.com/deepmind-media/gemini/gemini_3_pro_fsf_report.pdf)
- [Frontier Safety Roadmap](https://www.anthropic.com/responsible-scaling-policy/roadmap)
- [A small number of samples can poison LLMs of any size](https://www.anthropic.com/research/small-samples-poison)
- [Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned](https://arxiv.org/abs/2209.07858)
- [Sabotage evaluations for frontier models](https://www.anthropic.com/research/sabotage-evaluations)
- [Core Views on AI Safety](https://www.anthropic.com/news/core-views-on-ai-safety)

[All units in Learn to use AI](https://rabbithole.pro/courses/learn-ai)

Source page: https://rabbithole.pro/courses/learn-ai/ai-safety