Let me open by thanking the Spanish EU presidency for their formal support and collaboration on this very important event.
There is a diverse range of civil society organisations here today, with a wealth of expertise, knowledge and lived experience that can help achieve our common goal of a strong, effective piece of legislation that protects the rights of all. As diverse a group that there is here today, we as human rights advocates are united in what the prioritise should be as we face the final weeks of negotiations of this landmark law.
We need:
A truly human rights-based approach, a holistic framework that ensures that we priortise the protection of the most vulnerable and at-risk groups.
Strong, accountability and transparency measures.
Strict regulation, and prohibitions as necessary on the use of AI in the context of law enforcement and migration.
Access to effective remedy and redress.
The need for a human rights-based approach
Today, at a crucial juncture in negotiations we are hosting a dialogue between civil society and member state representatives on the EU’s AI Act. The stated goal of this ambitious law is clear: to protect people, their rights and their wellbeing in the face of the increasing use of AI to make life-changing decisions such as; whether you get that job; whether you and your family get awarded social support you desperately need; which school your child gets accepted to; whether or not your asylum claim and lifeline to safety is approved; or whether you could find yourself wrongly categorised as a criminal. The list goes on – the stakes are dangerously high. Yet, this very fundamental goal of protecting people is at peril unless we get this keystone piece of legislation right.
For that reason, we appeal for a holistic framework to protect human rights in the AI Act. Such an approach would necessitate going beyond existing legislation. The GDPR, as important a piece of the puzzle as it is, is not preliminary designed nor can be adequately effective at preventing and mitigating discrimination in AI decision making for instance.
The use of AI systems for law enforcement and migration control
Furthermore, a human rights-based approach, by default would ensure that the most vulnerable, marginalised and at-risk enjoy additional protections. This is why the numerous proposals for exceptions and loopholes to the duty to register high risk uses of AI in law enforcement and migration control should be of concern to all of us.
The use of facial recognition by law enforcement and immigration authorities poses a particular threat to human rights because it can result in unjustified deprivations of liberty, and has shown repeatedly through extensive research to amplify and exacerbate discriminatory outcomes. This is why it will be of vital importance from a rights-protecting perspective that the AI Act prohibits mass surveillance use of biometric surveillance technology such as facial recognition.
The introduction of strong accountability and transparency measures
A persistent challenge when it comes to accountability for AI decision making is transparency. Understanding, how to use what data, in what ways that decisions are made. This is particularly important when it comes to enforcement of equality laws where the burden of proof shifts to the authority that makes a decision to disprove discrimination has occurred. How can an individual challenge such a decision if the user of an AI system to make that decision does not themselves understand it. Human rights impact assessments, engagement with civil society, and transparency obligations regarding the use of AI will be crucial to ensuring that the AI Act actually succeeds in its ambition to ‘see into the black box’.
Empowering people through redress mechanisms
Good well thought out legislation matters, and can impact entire generations. Ireland’s entry into the EU, thanks to equality legislation, led to the abolition of a marriage bar for women in public service jobs. Without the lifting of that ban my own mother would have had to give up her job. To protect individuals and indeed communities, any laws worth the paper they are written on must necessarily include an effective remedy and clear actionable rights. That is why the redress mechanisms within the EU AI Act, and how they interact with existing equality and privacy laws will be of paramount importance.
Conclusion
I would also like to thank our co-hosts at the Open Government Partnership and Democracy Society for joining with us to realize this shared goal of transparent, participatory, inclusive and accountable governance, and thanks once again our partners at the Spanish EU Presidency for their commitment to making this diagloue happen.
We are fortunate to have so many leading experts around our real and virtual roundtable today, from government, and civil society; we very much look forward to the expert discussions. Thank you all for being here and for your commitment to this ambitious project.
Let this be another conversation, in the series of many.
British Public Oppose Secret Surveillance Powers and Want Strong Protections for Encrypted Communications
CDT released nationally representative polling data revealing that a the British public are broadly concerned about the security, privacy, and free speech consequences of allowing law enforcement to access encrypted communications.
General-purpose AI Code of Practice Implementation: A Rights Blindspot
CDT Europe's analysis of the entry into application of the European Commission’s enforcement powers towards general-purpose AI (GPAI) models, detailed and operationalised in the GPAI Code of Practice.
As Brussels starts emptying for the summer, the risks of frontier AI models continue to capture the attention of policymakers as a key chapter of the AI Act becomes enforceable, and disclosure rules around the use of AI systems and AI-generated content are further clarified.
CDT Europe’s Feedback on the Draft Guidelines for the Classification of High-Risk AI Systems under the AI Act
CDT Europe responded to the European Commission’s consultation on the draft guidelines for the classification of high-risk artificial intelligence systems.
