RSF’s five proposals for a European AI plan that protects the right to information

The publication of the AI Act’s final version when it is formally adopted on 22 April must mark the start of a more ambitious European plan for AI and the right to information. This law is necessary and foundational, but it fails to address the challenges in the information area, an omission fraught with danger for democracies. Additional parliamentary work is needed to rectify the situation.

Continuing on the path established by RSF’s proposed 12-point New Deal for the Right to Information, RSF now calls for the development of a European AI plan that will enable the development of an AI industry that protects the right to information in EU member countries.

“Now that the AI Act has been translated into all of the languages of the EU member countries, the European AI industry is set to grow. But the European AI market currently lacks regulation to protect the right to information. To fill these gaps, we call on the European Commission to establish a European AI plan that safeguards information integrity and access to reliable and pluralistic information for European citizens – and at the same time acts as a driving force at the international level.”

Vincent Berthier
Head of RSF’s Tech Desk

The European AI plan that RSF proposes consists of five key recommendations that will force designers of AI systems to develop products that are safer and more effective in terms of respecting the right to information than those that are currently on the market.

RSF’s five recommendations for a European AI plan that protects the right to information: 

• Regulate the development of industrial standards for AI systems dedicated to information

  So that algorithms used in the information sector respect journalistic standards, RSF calls for the implementation of industrial standards based on the principles of the Paris Charter on AI and Journalism launched by RSF in July 2023.

• Make the systems used in the production of information subject to legal safety requirements

  In order to prevent the creation of content for disinformation purposes, AI systems that are made available on the market must provide solid guarantees in terms of the protection of journalistic sources and content transparency. This is also needed to prevent the proliferation of unidentified deepfakes that can be used, inter alia, to usurp the identity of journalists and discredit journalistic work. Furthermore, content producing systems should be trained to take account of accuracy, pluralism and diversity of points of view.

• Require the implementation of standards for the reliability of data used by AI systems 

  So that the media content used by AI systems complies with journalistic quality standards, RSF calls for reference to be made to reliability standards such as the Journalist Trust Initiative. The sources used to produce content must also be cited and the authors of the original data must be compensated. This remuneration should be done within the framework of collective negotiations with publishers and representatives of journalists.

• Fund and encourage the open data movement to enable AI systems to access more varied data 

  This aim of this measure would be to enable systems producing information content to have access to data allowing a diversity of opinions in the content produced

• Develop platforms dedicated to public documentation of AI systems

  Journalists and media must have access to reliable and independent impact studies, evaluations and recommendations regarding the AI systems that are available on the market.