The European Disability Forum (EDF) hosted a webinar on the 15th of October 2024 that focused on the impact of the EU’s Artificial Intelligence (AI) Act on persons with disabilities. The event introduced a new toolkit designed to help disability organisations engage in advocacy and monitor the national implementation of the AI Act. This new EU Regulation, which came into force on the 1st of August 2024, classifies AI systems by risk and imposes various safeguards. As AI technologies grow more pervasive in everyday life, their regulation has become critical, particularly for the disability community in ensuring accessibility and non-discrimination.
The AI Act: Safeguards and Risks for Autistic Community
Central to this law is the classification of AI systems into four risk categories: unacceptable risk, high risk, limited risk, and low risk. Systems that present an “unacceptable risk” to human rights are outright banned, while high-risk systems, such as those used in employment, education, and healthcare, are subject to regulations. This is especially important for autistic people, who may be affected by emotion-recognition AI systems where they are used to make critical decisions about their lives. The Act emphasizes the need for these high-risk systems to comply with accessibility standards and undergo bias testing.
One major concern highlighted in the webinar was the AI Act’s exceptions for law enforcement and immigration. These exceptions allow authorities to use facial recognition and other AI tools, which can misinterpret the behaviours of autistic people, such as avoiding eye contact or having different facial expressions. This poses a significant risk for autistic individuals being unfairly treated by AI systems used by police and immigration authorities.
Engaging with the AI Act: EDF’s Toolkit
- The EDF toolkit serves as a practical guide for organisations to influence national implementation of this EU Regulation which will enter into effect in stages. The first stage is Feb 2025 with the full act becoming applicable by August 2026.
It outlines how organisations can participate in the national implementation process, monitor the use of AI systems, and advocate for stronger safeguards. Key recommendations include providing feedback during public consultations and engaging with national authorities to ensure that AI systems are designed to meet the needs of autistic people.
A critical feature of the AI Act is the right to explanation, which allows individuals affected by high-risk AI decisions to understand how those decisions were made. Autism organisations can play a crucial role in raising awareness of this right and supporting autistic people in exercising it.
National Advocacy and Collaboration
While general AI systems will be largely regulated by EU policies, they will also be subject to codes of conducts. National disability organisations can participate in the development of these codes of conducts. To get involved organisations can contact the responsible ministry at national level that will draft the national code of conduct, and the consumers and digital rights’ organisations in their country.
The webinar also underscored the importance of disability organisations engaging with the ‘market surveillance authorities’, which will be responsible for enforcing the AI Act at the national level. These authorities must ensure AI systems meet accessibility and bias requirements, making it essential for autism organisations to establish strong relationships with them, and for example lodge complaints in case of violation of the Act. It is up to the member states to either appoint this role to an existing authority or create one.
Another consideration is Article 77 of the act which grants to national authorities and bodies protecting fundamental rights a right to access documentation produced under the AI Act where such information is relevant to their mandate. Member States must publish a list of their respective competent authorities or bodies by 2 November 2024.
The Way Forward
The toolkit developed by EDF will be regularly updated as new EU guidelines are released so we can stay informed on AI Act implementation. By staying engaged in legislative processes, monitoring AI systems, and building alliances, we can ensure that AI technologies enhance inclusion and respect the rights of autistic people. As AI becomes increasingly integrated into key areas of life, the active involvement of autism organisations will be critical in shaping a fair and accessible future for all.
Further, the European AI & Society Fund has launched two funding calls for non-profits. The first focuses on implementing the AI Act in EU countries or at the EU level, with a deadline of 10th November 2024. The second targets AI system accountability through testing or auditing to ensure fairness.