We commend the Commission's effort to clarify legal responsibility and accountability for AI applications and systems. We agree with the Commission's view that, in principle, "each obligation should be addressed to the actor(s) who is (are) best placed to address any potential risks". However, complex situations might arise such as when AI systems are integrated into a product made by a third party, and/or receive continuous software updates. For instance, if large consumer businesses were not confident that suppliers providing the AI subcomponents of their products would bear appropriate responsibility in case their algorithm's failure resulted in the end product harming consumers, they might decide to bring most AI work in-house to better control risk. This could reduce competition and innovation, and particularly affect SMEs. We would therefore suggest that the Commission further explore how such legal clarity and confidence can be achieved.