That's a really important question, and it's one of the major risks that we think needs to be addressed around facial recognition use.
I'll come back to what I said before in terms of the internal safeguards we've mentioned. One of the most important is the testing piece. We make sure that we are opening up our facial recognition system to independent third party testing to make make sure that we are training and testing it ourselves in such a way that we are confident it is performing accurately and in a way that minimizes gaps across different demographic groups.
I would really like to emphasize as part of my contribution today that the testing piece is a really important part of making sure that technology is performing in an accurate manner, which it can do. It's made incredible strides in the best-performing algorithms in recent years, but there are many algorithms out there that aren't as accurate. You need to be able to test them to make sure that when, for example, police are using them, they are using the most accurate systems.