ITG NowHome › AI glossary › AI alignment
What is AI alignment?
AI alignment is the work of making an AI system do what people actually want and follow human values.
- Category Safety & law
- Also called alignment
- Updated 20 Oct 2026
AI alignment, explained simply
Alignment means the AI's goals match yours. A model that is aligned tries to be helpful and honest, and avoids harm.
It is harder than it sounds. People don't always say exactly what they mean, and a model can follow words but miss the point.
Companies work on alignment by training models with human feedback. You see the result when a chatbot refuses a harmful request.
On ITG Now, our daily AI news stories sometimes cover how companies try to make their models behave well.
Related terms
AI safetyGuardrailsRed teamingTrainingSuperintelligenceAGI
Where you'll see it on ITG Now
Questions people ask about AI alignment
What does an aligned AI mean?
It means the AI tries to do what you really want and follows human values. It also avoids harmful actions.
AI alignment vs guardrails: what is the difference?
Alignment is shaping what the model wants to do. Guardrails are extra checks that block unwanted results.
Is alignment solved?
No. It is still an active area of research, and models can still make mistakes.