An international AI Treaty? Council of Europe AI Convention, an update

The draft text has been published for an international Convention on AI focussed on human rights, the rule of law, and democracy which is being developed by the Council of Europe (CoE), an international human rights organisation. The AI Convention, sometimes referred to as the AI Treaty, is intended to be the first legally binding international convention on AI.  EU States and other countries – including the UK and US – could be signatories.

We previously provided an overview here before the text was available. Now the text is available here, we pick out some of the key points below.  Those who review the full text here will see text in black (considered agreed) and red (for further discussion); clearly much has already been agreed but further work is required. 

• What are the objectives of the Convention?

According to the draft text:

1. The provisions of this Convention aim to ensure that activities within the lifecycle of artificial intelligence systems are fully consistent with human rights, democracy and the rule of law. 
2. Each Party  [i.e. country which signs the Treaty] shall adopt or maintain appropriate legislative, administrative or other measures to give effect to the provisions set out in this Convention. These measures shall be graduated and differentiated as may be necessary in view of the severity and probability of the occurrence of adverse impacts on human rights, democracy and the rule of law throughout the lifecycle of artificial intelligence systems. This may include specific or horizontal measures that apply irrespective of the type of technology used. 
3. In order to ensure effective implementation of its provisions by its Parties, this Convention establishes a follow-up mechanism [including future conferences of the Parties and agreement to international co-operation].

• How is AI defined? AI has the same definition as that adopted by the OECD and the EU in the EU AI Act. This convergence is notable, as it was understood previously that the Treaty and AI Act differed in how they intended to define AI.
• What is the scope of the Convention?  The draft text includes various options for further discussion, in particular around the extent to which research and development, national security and national defence should fall outside of the Convention. 
• What are the Obligations on each Party (i.e. signatory country)? Each Party shall adopt or maintain measures to protect human rights, which maintain the integrity of democratic processes and respect for rule of law and which: 
  • respects human dignity and individual autonomy;
  • ensures transparency and oversight;
  • ensures accountability and responsibility;
  • ensures equality and non-discrimination;
  • ensures privacy and personal data protection;
  • preserves health and, subject to discussions, the environment;
  • promotes reliability and trust; and
  • encourages safe innovation.
• What remedies are required for individuals who are caused loss by an AI system?  the draft text is that each Party shall take measures ensuring the availability of accessible and effective remedies, including ensuring sufficient and proportionate documentation / information is retained to allow the affected persons to contest the decisions, and also that there are appropriate procedural safeguards in place. 

What's next? It is not known by when the Treaty is expected to be agreed.  However, there is growing appetite within and across jurisdictions for AI-specific policy, regulations and co-operation - such as the Bletchley declaration, EU AI Act and UK's White Paper on AI regulation - so it is safe to assume there is likely political momentum behind an AI Treaty.  Whether and to what extent national laws will need amending should a country sign-up to the Treaty will depend on that country, but it does raise the possibility of new / updated laws as well as both (additional) national and international scrutiny of those laws.

If you would like to discuss how current or future regulations impact what you do with AI, please contact Tom Whittaker, Brian Wong, David Varney, Lucy Pegler, Martin Cook or any other member in our Technology team.