Category
Keeping AI systems safe and accountable
Every layer of the AI stack can fail, be attacked, or be misused - and every serious team plans for all three. These entries cover the safety and governance layer: what AI safety and alignment actually mean, how attacks like prompt injection and jailbreaking work and how teams defend against them, and the practices and rules - responsible AI programs, model cards, and regulation - that keep AI systems accountable to the people who use them.
Entries in this category
- AI regulation Laws and frameworks that govern how AI systems can be built and deployed, usually scaled to the risk they pose. Read the entry
- AI safety The practice of preventing AI systems from causing harm, from everyday failures to large-scale misuse. Read the entry
- Alignment Training AI models so their behavior matches human intent and values, not just the literal words of an instruction. Read the entry
- Jailbreaking Crafting prompts that talk an AI model into ignoring its safety training and producing restricted output. Read the entry
- Model cards A structured document describing an AI model's intended use, training data, evaluations, and known limitations. Read the entry
- Prompt injection An attack where malicious instructions hidden in content an LLM reads override what the developer told it to do. Read the entry
- Responsible AI The organizational practice of building AI systems that are fair, transparent, and accountable. Read the entry
Where this layer fits
Safety work cuts across the whole stack rather than sitting at one level. It starts inside the model with the training techniques covered in core concepts, gets enforced at the application layer where agents add guardrails and human oversight, and gets measured by evals like red-teaming that probe whether the defenses actually hold.
Not sure where to start? AI safety is the entry that frames everything else in this category. Or browse everything at once in the A-Z index.