The Software & Information Industry Association (SIIA) has submitted its response to the European AI Office’s consultation on creating a Code of Practice (CoP) for trustworthy general-purpose AI (GPAI). Representing nearly 400 companies across industries like education, financial services, and software, SIIA emphasizes the importance of transparency, urging the CoP to promote practices that are both effective and aligned with existing industry standards. SIIA advocates for the use of model cards to improve transparency while acknowledging the complexities of downstream AI use. They caution against overly rigid transparency requirements that might expose trade secrets or security vulnerabilities, and suggest the CoP incorporate flexibility to adapt to ongoing developments in technical standards.
SIIA also stresses the need for the CoP to align with international best practices in risk assessment, including frameworks from the National Institute of Standards and Technology (NIST), the G7 Hiroshima AI Code of Conduct, and the emerging AI Safety Institute Network. The association recommends building the CoP on existing standards to reduce the compliance burden on companies while supporting global innovation and interoperability. Additionally, SIIA encourages a risk assessment framework that evolves with advancements in AI measurement and standards, ensuring both responsible development and continued technological progress.
