Safe

Safe

Evaluate Claude and Codex outputs to master AI safety decisions.

Chapter 1

The Specialized AI Tools

Chapter 2

Mastering Different AI Skills

Chapter 3

Safety Ethics and Training

Ch 3 · Safety Ethics and Training
federico

federico Claude AI Safety Ethics

Constitutional AI! Claude learns ethical rules, to stay true and kind!

Jul 15

federico

federico Claude follows a 'Constitution'—a set of rules that tell it to be ethical and helpful without needing constant human intervention.

Jun 13

federico

federico Reinforcement Learning from Human Feedback (RLHF) uses human testers to rank AI answers, teaching models like Claude what we find helpful.

Jun 13

federico

federico Because Codex learned from public code, it can accidentally repeat common mistakes or security flaws found in human-written software.

Jun 13

federico
Safe
BETA

federico Evaluate Claude and Codex outputs to master AI safety decisions.

Jun 12

federico

federico Developers work to ensure models like Claude admit when they don't know an answer, rather than making up false information.

Jun 12