OpenAI improves coding for low-resource languages

OpenAI's blog explained how the company is improving ChatGPT across low-resource programming languages, with new evaluations for software engineering, SQLite, and COBOL migration. The effort addresses a gap that has shaped AI coding tools: they tend to perform best in popular languages such as Python and JavaScript, while legacy and niche languages receive less attention. That leaves important workloads underserved. Many enterprises still run critical infrastructure on aging systems, including COBOL applications, and modernizing those systems is difficult and expensive. Better AI support could help teams understand legacy code, plan migrations, and reduce the risk of introducing errors. OpenAI's new evaluations are meant to measure progress in these less-covered areas, including software-engineering tasks and database work. The company says it is working to improve ChatGPT's performance across low-resource programming languages, though the blog frames the work as ongoing rather than complete. The announcement matters because AI coding tools are becoming part of professional workflows, and trust depends on reliable performance beyond the most common languages. If AI assistants can help with COBOL and other legacy environments, they could support maintenance and modernization efforts that are often slow and under-resourced. At the same time, migration work carries high stakes, so any AI-generated code needs human review and strong testing. OpenAI's update is a sign that coding AI is moving toward broader coverage, but low-resource languages will remain a useful benchmark for how well these tools handle real-world software.