Brazil’s B3 adopts Cursor AI coding tool for critical market systems

When a national stock exchange declares that AI code assistants are now part of its standard development toolkit, it is a milestone in the normalization of AI‑authored code in safety‑critical infrastructure. B3’s move shows that even highly regulated entities now see the productivity gains from tools like Cursor as worth the governance overhead.

Strategically, this accelerates diffusion of AI fluency into the financial system’s plumbing. Thousands of small changes to trading platforms, clearing systems and risk engines will now be at least partially machine‑generated. If managed well, that can improve quality and speed; if not, it risks embedding subtle bugs and vulnerabilities at the very core of a country’s capital markets. B3’s emphasis on keeping human review and existing QA processes is a recognition that “AI in the loop” is the only politically acceptable way to do this at scale right now.

For the AGI race, this is another example of how agentic tools will quietly permeate critical infrastructure years before fully autonomous systems are politically viable. The more exchanges, banks and utilities bake these tools into their workflows, the harder it will be to unwind AI dependencies later, which effectively locks in demand for ever‑more capable models.