How do you handle client info when using ai tools at work?
Something's been bugging me lately and I want to know if others deal with this too. when you paste work documents into chatgpt or claude , what do you actually do about sensitive client info? Like names, case numbers, dates, anything identifying? do you manually go through and delete everything before pasting? or honestly just paste and hope for the best?
the manual delete thing is annoying as hell, miss one thing and you've potentially leaked client data into some company's training pipeline. wondering if a tool that automatically scrubs that stuff locally before you copy anything into an ai and lets you swap the real info back after, would actually be useful. or is this a non-problem that people have already figured out some other way?