Human translation vs. AI in contracts: where the real risk lies

It's reasonable to use machine translation to understand what a document is about. It's risky to use it, without human review, to produce the version the parties will sign or file with an institution. The difference isn't in the fluency of the output, which is usually good: it's in where the tool gets it wrong without it being obvious at a glance.

Where it fails most often

• Modal verbs with legal effect: confuses "shall" (obligation) with "may" (discretion)
• False-friend legal terms: translates "warranty" as a product "guarantee," not a contractual representation
• Contractual definitions: doesn't keep a term defined in clause 1 consistent throughout the rest of the document
• Cross-references between clauses: loses the number or order when restructuring a sentence
• Proper names and figures: in low-quality scanned documents, it can alter an amount or a date without the error being visible

Where it works reasonably well

For a first orientation read, to get the general sense of a long document before deciding whether it needs an official translation, or as a starting draft that's later reviewed and corrected term by term, machine translation is a legitimate and useful tool. The problem isn't the tool: it's using it as a final product without human review and legal judgment.

That's why we offer proofreading and editing of machine translations as a standalone service: there's no need to translate a document from scratch if a draft already exists — just someone with legal judgment to review it before it's signed or filed.