AI-assisted migration offers a path through the COBOL backlog
Banks, airlines, government agencies, and other organizations still depend on decades-old COBOL systems, making updates risky and expensive. OpenAI says its latest AI models can make “substantial improvements” in converting legacy code into modern programming languages, with the goal of helping institutions maintain and modernize software they cannot afford to replace all at once.
If the system performs reliably at scale, AI-assisted migration could reduce some of the manual work involved in translating and testing old code. That would be particularly valuable where small changes to established systems can have large operational consequences. The announcement nevertheless frames the capability as a company claim, and its practical value will depend on how accurately the models translate code, how well organizations can verify the results, and whether automated migration introduces new defects into systems that are already difficult to change.