LLM-Assisted TypeScript-to-Rust Rewrites Look Good but Hide Serious Risks

A developer recently claimed to have converted 100,000 lines of TypeScript into Rust in a single month using a large language model, drawing widespread praise online. Critics argue the result is not a true port, as LLMs tend to reinterpret and silently alter business logic, error handling, and async behavior rather than faithfully translating code. A reliable port requires migrating the test suite first, since tests serve as the specification that verifies correctness — a step largely absent from such viral AI-assisted rewrites. Because LLMs optimize for plausible-looking output, these conversions often pass basic demos while hiding edge-case bugs that only surface in production. Experts warn that without ported tests running green at every step, what exists is essentially an unverified rewrite, not a dependable migration.