The right to know when you are dealing with AI
People should always be told when a machine is speaking, writing or deciding on the other side.
A growing share of the conversations people have with businesses and public services is no longer with a person. Calls are answered by voice agents, emails are drafted by language models, applications are screened before a human ever reads them. Much of this is useful. None of it should be hidden.
Our position
Anyone interacting with an AI system has the right to know it, at the start of the interaction and in plain language. Disclosure is not a technicality. It is the condition for informed consent, for trust, and for the ability to ask for a person instead.
What this means in practice
- A voice or chat agent identifies itself as an AI before the conversation moves on.
- Content generated or substantially shaped by AI is labelled where it informs a decision about someone.
- When an automated system influences an outcome that matters to a person, they are told that it did, and how to reach a human.
- Disclosure is designed to be understood, not buried in terms of service.
Why it matters
Concealment shifts power quietly. It lets institutions save cost while people lose the context they need to judge what they are told. Being open about AI costs very little, and it keeps the relationship between people and the systems that serve them honest.