OpenAI announced substantial improvements in its latest models regarding the migration of aging software written in older programming languages into modern code. The company highlighted COBOL as a specific target, noting that this capability could assist organisations with maintaining and modernising older systems. This development follows a broader industry concern where critical infrastructure, including banking and airline operations, relies on antique computer code that few programmers dare to modify. Both OpenAI and Anthropic have recently stated they are actively working on this specific technical challenge to address the looming shortage of skilled COBOL developers.
The actual significance lies in the potential reduction of manual effort required for legacy system updates, which often consumes a disproportionate amount of IT resources. This approach addresses the immediate risk of system failure without necessarily altering the underlying architecture of the software running on government or corporate networks. The focus remains strictly on the translation process rather than the operational behaviour of the applications themselves.
- Targeted specifically at COBOL and other low-resource programming languages
- Claims substantial improvements in migration accuracy for older systems
- Does not imply changes to customer products, accounts, or API behaviour
