1 00:00:00,120 --> 00:00:03,159 Speaker 1: Back to the Privacy Commissioner who's found that the facial 2 00:00:03,200 --> 00:00:06,880 Speaker 1: recognition trial by the food Stuffs people was compliant with 3 00:00:06,960 --> 00:00:10,160 Speaker 1: the Privacy Actor. It also worked, violent behavior was reduced 4 00:00:10,160 --> 00:00:13,720 Speaker 1: by sixteen percent, prevented more than one hundred serious harm incidents. 5 00:00:13,760 --> 00:00:17,400 Speaker 1: Michael Webster's the Privacy Commissioners with us morning. Good morning mate, 6 00:00:17,520 --> 00:00:19,880 Speaker 1: what sort of lens did you look at this through? 7 00:00:21,880 --> 00:00:24,400 Speaker 2: So when we looked at the trial of the use 8 00:00:24,440 --> 00:00:27,400 Speaker 2: of facial recognition technology, we looked at it from the 9 00:00:27,400 --> 00:00:31,000 Speaker 2: point of view about whether its use was was justified, 10 00:00:31,160 --> 00:00:36,360 Speaker 2: whether it was proportionate and necessary to address the problem, 11 00:00:36,400 --> 00:00:40,040 Speaker 2: the problem being serious harm caused through retail crime. And 12 00:00:40,080 --> 00:00:43,160 Speaker 2: we also looked at as it was used, what privacy 13 00:00:43,240 --> 00:00:45,919 Speaker 2: safeguards needed to be put in place to ensure that 14 00:00:46,440 --> 00:00:50,120 Speaker 2: as much as possible New Zealander's privacy was protected. 15 00:00:50,120 --> 00:00:53,920 Speaker 1: At the same time, did it pass easily or scrape through. 16 00:00:55,600 --> 00:01:00,480 Speaker 2: We determined that its use by Foodstuffs was compliant the Act, 17 00:01:01,320 --> 00:01:03,480 Speaker 2: but we also recommended to food Stuff so there's a 18 00:01:03,520 --> 00:01:06,679 Speaker 2: number of steps they need to take to improve and 19 00:01:06,800 --> 00:01:10,280 Speaker 2: enhance how it's being used, and we also set out 20 00:01:10,360 --> 00:01:13,680 Speaker 2: some recommendations for others who might be thinking of using 21 00:01:13,680 --> 00:01:18,240 Speaker 2: it as well. There are still concerns around technical bias 22 00:01:18,319 --> 00:01:22,520 Speaker 2: issues with the use of facial recognition technology software that 23 00:01:22,560 --> 00:01:24,120 Speaker 2: comes from overseas. 24 00:01:23,959 --> 00:01:27,280 Speaker 1: Right, is that improving rapidly? Do you know, technologically speaking 25 00:01:27,360 --> 00:01:27,560 Speaker 1: or not? 26 00:01:28,880 --> 00:01:31,600 Speaker 2: It is improving. What we would like to see here 27 00:01:31,640 --> 00:01:34,840 Speaker 2: in New Zealand is a development of our own data 28 00:01:34,959 --> 00:01:39,080 Speaker 2: set to use with FRT, so representative of the New Zealand. 29 00:01:38,840 --> 00:01:41,080 Speaker 1: Population that'll come with time, won't it. 30 00:01:42,800 --> 00:01:46,000 Speaker 2: We're certainly recommending to the powers that be that that 31 00:01:46,040 --> 00:01:46,880 Speaker 2: sort of thing happened. 32 00:01:47,120 --> 00:01:49,920 Speaker 1: Right when you say proportionate, who decides on proportion? 33 00:01:51,880 --> 00:01:55,280 Speaker 2: Well, what Food Stuff did was get an independent evaluator 34 00:01:55,320 --> 00:01:58,320 Speaker 2: and to examine the degree to which the use of 35 00:01:58,360 --> 00:02:04,680 Speaker 2: the technology reduced of say aggression or bullying behavior, or 36 00:02:05,320 --> 00:02:09,280 Speaker 2: large scale retail theft. And so we've examined that data. 37 00:02:10,160 --> 00:02:13,280 Speaker 2: We also went out and spoke to customers and staff 38 00:02:13,280 --> 00:02:16,440 Speaker 2: and supermarkets, looked at how the systems were being used 39 00:02:17,200 --> 00:02:19,639 Speaker 2: and given the degree of concern that's out there, Mike 40 00:02:19,680 --> 00:02:24,079 Speaker 2: about retail crime, we determined that the way in particular 41 00:02:24,120 --> 00:02:28,520 Speaker 2: that Foodstuffs had implemented the system met that very high 42 00:02:28,520 --> 00:02:30,160 Speaker 2: threshold in the privacy excess. 43 00:02:30,200 --> 00:02:32,359 Speaker 1: So how much weight did you place on so violent 44 00:02:32,400 --> 00:02:35,760 Speaker 1: behavior reduced by sixteen percent? Was that a slam dunk 45 00:02:36,160 --> 00:02:38,080 Speaker 1: or was that kind of will put that in with 46 00:02:38,160 --> 00:02:42,120 Speaker 1: the mix of other things we worry about. 47 00:02:42,240 --> 00:02:46,520 Speaker 2: That was an important statistic. What was also important was 48 00:02:47,080 --> 00:02:49,440 Speaker 2: some of the safeguards put in place. For example, Mike, 49 00:02:49,480 --> 00:02:52,800 Speaker 2: New Zealanders might be surprised to know that actually close 50 00:02:52,880 --> 00:02:57,000 Speaker 2: to two hundred and twenty six million phases were scanned 51 00:02:57,120 --> 00:02:59,960 Speaker 2: during the trial, but ninety nine point nine nine nine 52 00:03:00,320 --> 00:03:03,560 Speaker 2: seen to them were immediately deleted. And that was because 53 00:03:03,560 --> 00:03:06,480 Speaker 2: we suggested to Foodstuffs and they took on board the 54 00:03:06,520 --> 00:03:09,120 Speaker 2: idea that if there wasn't a match, the face get 55 00:03:09,120 --> 00:03:12,000 Speaker 2: immediately deleted. So throughout the life of the trial, only 56 00:03:12,040 --> 00:03:16,720 Speaker 2: about three thousand or so scans of interest were actually 57 00:03:16,800 --> 00:03:18,200 Speaker 2: actioned and fed into the system. 58 00:03:18,639 --> 00:03:21,160 Speaker 1: See, I, as a punter, couldn't care less. If you 59 00:03:21,160 --> 00:03:23,840 Speaker 1: can reduce crime and it makes life easier for them 60 00:03:23,880 --> 00:03:26,079 Speaker 1: to do business, I'm all for it. Am I normal 61 00:03:26,160 --> 00:03:27,000 Speaker 1: or not really normal? 62 00:03:29,200 --> 00:03:32,880 Speaker 2: We did a survey this year and we asked questions 63 00:03:32,960 --> 00:03:35,680 Speaker 2: about exactly that, Mike, and what we saw was that 64 00:03:35,760 --> 00:03:38,880 Speaker 2: about two thirds of New Zealanders we're willing to see 65 00:03:39,000 --> 00:03:43,480 Speaker 2: an increased use of what we described as privacy intrusive technology, 66 00:03:43,520 --> 00:03:47,320 Speaker 2: which FRT facial recognition technology is. We're willing to see 67 00:03:47,360 --> 00:03:53,800 Speaker 2: its use if it reduced theft, if it increased personal safety. 68 00:03:54,760 --> 00:03:57,920 Speaker 2: That said, forty to fifty percent of New Zealanders we're 69 00:03:57,920 --> 00:04:01,240 Speaker 2: concerned about the use of facial red condition technology as well. 70 00:04:01,320 --> 00:04:04,440 Speaker 2: So people are willing to see this technology being used, 71 00:04:04,440 --> 00:04:07,400 Speaker 2: but they want it to be used in a way 72 00:04:07,440 --> 00:04:10,520 Speaker 2: that is absolutely as privacy protective of them and their 73 00:04:10,520 --> 00:04:11,960 Speaker 2: personal information if possible. 74 00:04:12,000 --> 00:04:13,920 Speaker 1: All right, good, So I appreciated Michael Webster, who's the 75 00:04:13,960 --> 00:04:14,760 Speaker 1: Privacy Commissioner. 76 00:04:15,240 --> 00:04:18,159 Speaker 2: For more from The mic Asking Breakfast, listen live to 77 00:04:18,240 --> 00:04:21,320 Speaker 2: news talks. It'd be from six am weekdays, or follow 78 00:04:21,360 --> 00:04:22,880 Speaker 2: the podcast on iHeartRadio.