AI chatbots in Sweden must identify themselves from August 2

From August 2, people in Sweden must be told when they are dealing directly with an artificial-intelligence system, unless that is already obvious.

The EU's new transparency rules cover chatbots used by websites, employers and public bodies. Providers must also add machine-readable marks to synthetic audio, images, video and text so verification tools can identify AI-generated or manipulated material.

Visible labels are required for deepfakes. AI-written text about matters of public interest must also be disclosed unless a human has reviewed it and a person or organisation accepts editorial responsibility.

Employers and other operators using emotion recognition or biometric categorisation must inform the people exposed to those systems.

The European Commission says breaches of these transparency duties can bring fines of up to €15 million or 3% of worldwide annual turnover. The higher €35 million ceiling applies to prohibited AI practices, not these labelling rules.

A limited grace period until December applies only to the machine-readable marking obligation for some systems already on the market. It does not postpone the August 2 chatbot disclosure rule.

The rules give users a simple right: an AI system cannot quietly present itself as a person.

DailySweden

DailySweden's desk of editors, reporters, researchers and investigative journalists. We bring readers factual, truthful and objective reporting on the issues that matter to them.

Contact the editorial desk ›