We have written a number of articles, covering the hard position that the American judiciary has taken, in disciplining attorneys who filed briefs that contain bogus or fabricated citations to legal authorities, meaning case law, and secondary legal references. Lawyers have drawn as much as $10,000 fines for their carelessness, irrespective of the reason they allow AI research to creep into their court filings.
Compliance officers, as we have seen, have been supplementing their due diligence and enhanced due diligence queries with AI searches, and using the results tom decide on whether to approve (onboard) new bank clients, and some are not taking the additional, and necessary, step of actually verifying what the AI program has told them about their new customer.
Here's why this is risky, from a professional ( malpractice/negligence) viewpoint. Unlike lawyers, whose legal citation are often reviewed for accuracy by opposing counsel, and the presiding judge, A-vetted compliance decisions simply go into a file, and generally are never seen by anyone else. They do, however, surface when the customer is later found to be laundering money through his accounts, or is indicted, and a subpoena finds its way into the bank. of course it is far too late at that point to fix the error. If you don't check your AI-generated results in compliance, they may come back to haunt you.
One final and important point as to why you should be concerned; some open AI is continually learning from the data fed into it, and if a prior user input, or generated, bogus data on your target, you may very well receive that false information in your subsequent query. Are you sure you are using a Closed AI program? And what about security flaws possibly present in Open AI? you may have unwittingly further increased your chance of picking up erroneous results.
Bottom line, check everything that AI gives you back, Mr. or Ms. compliance officer.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.