All AI systems must be able to be traced back to the deployers. This ensures a level of accountability on the deployers of AI systems as it ensures all systems are, to their knowledge, safe and transparent. 

In addition to this, all deployers of AI systems must implement a risk management approach. With this, it maintains a rigid level of risk  management on every AI system. Any risks identified throughout this process can then be identified and resolved to ensure its safety for users. 

The OECD provide further guidance on the meaning of ‘accountability‘ as the expectation that organisations or individuals will ensure the proper functioning, throughout the entire lifecycle, of the AI systems that they design, develop, operate or deploy in accordance with applicable regulatory frameworks.  This then means deployers are accountable for any system they deploy, and its effects on society. One key point to mention here is the reference to the entire lifecycle of an AI system. The principle requires deployers and developers to continuously  regulate AI systems from the moment they are deployed to the moment they are ended. 

Documenting the process of developing these AI systems is essential to meeting the principle of Accountability. Through documenting every stage this provides hard evidence on the risk management and safe procedures taken in the lead up and throughout the entire lifecycle of an AI system.  Furthermore, documentation is often required through regulations to specific AI regulatory bodies within your jurisdiction. This acts as a layer of accountability as it ensures all necessary procedures and steps were taken, protecting yourself and your organisation. 

As you can see, accountability is essential in AI regulation as it ensure no anonymous harm can be done. Instead it assigns a clear level of responsibility for all AI outcomes meaning penalties can be provided were necessary.