Aviation certification regulators are on track to adapt and enhance the current regulatory framework.

 AI, and more specifically the machine learning (ML) field of AI, bears enormous potential for developing applications that would not have been possible with the development techniques used so far. The deployment of AI technology has therefore become a strategic priority for the aviation industry, including for safety-related applications. The ML integration represents a paradigm shift from traditional rule-based systems to those that learn behavior through a data-driven approach. As a result, ML models exhibit a probabilistic nature, inferring the most likely outcomes for new data based on relationships learned from training datasets. This results in additional difficulties in the certification andqualification process of AI-based systems, as they do not behave like traditional systems. Aviation certification regulators are on track to adapt and enhance the current regulatory framework. For example, the European Union Aviation Safety Agency (EASA) has been proactive, having prepared its AI roadmap for aviation, published in 2020. This effort includes the publication of a Concept Paper outlining objectives that will form the basis of a new regulation, called Part-AI. Similarly, the Federal Aviation Administration (FAA) initiated the development of its own AI roadmap in 2024, reflecting a growing recognition of the need for updated regulatory approaches to accommodate AI advancements.  On the other hand, the aviation industry has gathered to prepare a new standard for the development and certification of AI-based systems. The joint international committee, named SAE G34/EUROCAE WG-114, has already published two documents (AIR6988/ER-022 AI: Statement ofConcerns and AIR6987/ER-027 AI: Taxonomy), and the standard itself (ARP6983/ED-324) is expected to be published in 2026. The goal is to have it recognized by the civil aviation authorities as an acceptable means of compliance to their regulation.