Presentation by Director Hanne Juncher, Council of Europe
Thank you for inviting the Council of Europe to speak at this important event.
The Council of Europe is a regional intergovernmental organisation with 46 member states. It is quite distinct from the European Union.
We work across three pillars: human rights, democracy and the rule of law, addressing a vast range of issues from education to cybercrime. Our treaties are open to accession by non-member states, and we work increasingly with states from other regions, including Africa.
Our starting point is that AI comes with great promise, but there is also another side where it can be become a tool of political repression, undermine democratic processes, and result in serious human rights violations.
In 2022, the Committee of Ministers of the Council of Europe decided to open negotiations on a legally binding instrument on AI.
The negotiations involved the member states of the Council of Europe, 11 non-member states, and international and regional organisations, including the UN, the OECD, and the EU, as well as 68 representatives of civil society, academia, and industry. All the G7 states took part, and 28 out of the 38 OECD member states.
We did not have any African countries in the room, although we were in touch with several. I will come back to that.
The new Framework Convention on AI and Human Rights, Democracy and the Rule of Law was opened for signature in September. Ten states have already signed it, including the US, plus the European Union on behalf of the 27.
It will enter into force after the fifth ratification and then be open for accession also by states that did not take part in the negotiations.
The Convention is centred around seven basic principles:
- Human dignity and individual autonomy
- Transparency and oversight
- Accountability and responsibility
- Equality and non-discrimination
- Privacy and personal data protection
- Reliability, and
- Safe innovation
The Council of Europe, and all the states from other regions that negotiated the Convention, were committed to creating global dialogue on AI and having a treaty that will contribute to bridging the digital divides.
The treaty adopts a risk-based approach with parties obliged to provide remedies and procedural safeguards, and to set up a risk and impact management framework. The text uses the OECD definition of an AI system, and the treaty applies to both public and private sector use of AI.
A Conference of the Parties will oversee the implementation of the Convention and provide for its interpretation, as well as other forms of international cooperation and exchange of information.
Importantly, the Convention does not apply to trivial AIs, but only comes into play for AI solutions that may pose a risk to human rights, democracy or the rule of law.
This could for instance concern an administrative or judicial decision by a public authority impacting on the rights of an individual, or the decision by a private entity to recruit, to grant a credit, or to provide insurance cover.
Given the versatility of AI systems, there are many such situations where an individual may be facing an AI system taking, or informing, decisions impacting their life in a significant way.
And there is a very real risk that such decisions may, inadvertently, turn out to be biased and constitute unlawful discrimination, if the processing by the AI system is not properly managed and controlled.
This is where the Framework Convention adds value - by obliging its Parties to guarantee the same legal standards of protection in a case where an AI system is being used, as in any other case.
We are preparing a methodology for the risk and impact management, called HUDERIA. It should be ready later this year. It will be a non-legally binding practical tool that provides guidance to the Parties in a very detailed way.
I should add that the Framework Convention does not create new human rights specific to AI, but it ensures that all existing human rights will prevail, also in situations where an AI system is involved in taking, or informing, decisions.
When it comes to ensuring the safe use of digital technologies, it seems clear that purely regional approaches will not be enough. Having even access to these technologies is an issue globally, and so is protection against their abuse or misuse.
For that reason, the Framework Convention has not been designed for a particular regional human rights framework or mechanism. It is based on globally applicable norms, and also leaving the necessary space for adapting to the legal set-ups and needs of the Parties.
As I mentioned, States from the Americas, the Middle East and the Asia-Pacific region participated in the negotiations. Regrettably, we did not manage to have African states onboard before the negotiations ended on the 14th of March this year.
But without the active involvement of African states there can be no global governance of AI, and we want to reach out to all interested African states and regional organisations.
Last week we received a request from an African country for observer status with the AI Committee which is responsible for the Convention. The MFA described it as the first step towards accession. I cannot name the country because the procedure is under way, but we are delighted with this development.
Once the Framework Convention has entered into force, it will be open for accession by all like-minded countries.
I invite interested states to join up with us as soon as possible. My colleagues and I are at your entire disposal for questions related to the treaty and how to accede to it.
The most important work is the actual implementation of the Framework Convention, and that is all still ahead of us. I really hope you will join us.
Thank you.