AI safety
Research and practice to stop AI systems causing harm, now or as they get more powerful.
For example
Testing whether a chatbot gives dangerous instructions before releasing it.
Related terms
- Accountability Being responsible for what happens.
- AI alignment Making sure an AI's goals and behaviour match what people actually want.
- Balanced data Examples that fairly cover every group.
- Bias When an AI is regularly wrong or unfair for one group.
- Block and report Stopping contact from someone and telling the platform about harmful behaviour.
- Bystander Someone who sees something harmful happen and chooses to act, or not to.