Submission to the Department of Industry, Science and Resources on Safe and Responsible AI in Australia
4 October 2024
PWDA welcomed the opportunity to respond to the Australian Government's proposals paper on 'Safe and Responsible AI in Australia'.
This submission looks at the proposed regulatory framework for artificial intelligence (AI) from a disability rights perspective. The submission includes recommendations on how to regulate the use of AI and shows areas that may require further thought.
Key points:
- AI has significant implications for people with disability, including potential benefits and risks.
- The proposed principles for defining high-risk AI are comprehensive but require stronger consideration of accessibility and intersectionality.
- PWDA supports the introduction of mandatory safeguards for high-risk AI systems, with additional provisions for accessibility and assured individual rights.
PWDA urges the government to prioritise the rights and needs of people with disability in the development of AI regulation, ensuring that technological advancements promote inclusion and equality.
Recommendations:
Recommendation 1 - Adopt a Whole-of-Economy Regulatory Approach.
Recommendation 2 - Mandate accessibility requirements for AI systems.
Recommendation 3 - Prohibit certain AI uses, such as social scoring and non-consensual emotion recognition.
Recommendation 4 - Enhance human oversight and accountability measures.
Recommendation 5 - Strengthen informed consent and data protection provisions.
Recommendation 6 - Ensure inclusive representation and co-design with people with disability in the AI development processes.
Recommendation 7 - Organisations must conduct and publish disability impact assessments for high-risk AI systems before deployment and undergo ongoing monitoring.
Recommendation 8 - Establish funding streams for disability-led AI research and development.
Recommendation 9 - Develop and implement AI literacy programs specifically for the disability community.
Recommendation 10 - Establish accessible and effective complaint mechanisms for individuals to report issues with AI systems, including concerns about accessibility or discrimination.
Recommendation 11 - Actively engage in international efforts to develop harmonised AI standards and regulations that prioritise disability rights and accessibility.
Recommendation 12 - Use the CRPD as the framework underpinning the regulation of AI.
