1
00:00:04,400 --> 00:00:07,800
Speaker 1: Welcome to tech Stuff, a production from I Heart Radio.

2
00:00:12,039 --> 00:00:14,960
Speaker 1: Hey there, and welcome to tech Stuff. I'm your host,

3
00:00:15,080 --> 00:00:18,360
Speaker 1: Jonathan Strickland. I'm an executive producer with iHeart Radio and

4
00:00:18,440 --> 00:00:21,240
Speaker 1: I love all things tech and it is time for

5
00:00:21,320 --> 00:00:26,720
Speaker 1: the tech news for Thursday, March eleventh, twenty one. Let's

6
00:00:26,720 --> 00:00:29,600
Speaker 1: get to it. Last week I told you about how

7
00:00:29,640 --> 00:00:33,480
Speaker 1: the social network site gab, known for the right wing

8
00:00:33,560 --> 00:00:37,839
Speaker 1: political philosophy of most of its users, was the victim

9
00:00:38,000 --> 00:00:41,919
Speaker 1: of a data breach as a hacker accessed gab systems

10
00:00:42,159 --> 00:00:46,920
Speaker 1: and stole around seventy gigabytes worth of data, including three

11
00:00:47,200 --> 00:00:52,480
Speaker 1: million private posts. Well earlier this week, gab has been

12
00:00:52,680 --> 00:00:58,080
Speaker 1: hacked again and even went offline temporarily as the administrators

13
00:00:58,120 --> 00:01:02,200
Speaker 1: of the site conducted and investigate into the security vulnerability

14
00:01:02,360 --> 00:01:06,119
Speaker 1: that made this possible. Now, during the downtime, users who

15
00:01:06,120 --> 00:01:09,399
Speaker 1: are trying to go to gab got an error message,

16
00:01:09,480 --> 00:01:12,400
Speaker 1: but the site is back up as of this recording.

17
00:01:12,920 --> 00:01:16,959
Speaker 1: A hacker using the handle captain Jack Sparrow that's j

18
00:01:17,319 --> 00:01:21,039
Speaker 1: A x p A r Oh claims responsibility for both hacks,

19
00:01:21,600 --> 00:01:24,720
Speaker 1: though who knows if that handle represents just one person

20
00:01:24,920 --> 00:01:27,720
Speaker 1: or it's being used by a group. Working together. The

21
00:01:27,760 --> 00:01:33,280
Speaker 1: hacker used authentication tokens that they gathered during the first hack,

22
00:01:33,959 --> 00:01:36,240
Speaker 1: and they used those to carry out the second one.

23
00:01:36,280 --> 00:01:39,760
Speaker 1: They were able to access the systems because those authentication

24
00:01:39,800 --> 00:01:43,560
Speaker 1: tokens were still good, and that showed the GAB failed

25
00:01:43,600 --> 00:01:47,360
Speaker 1: to reset those security tokens. The hacker also left a

26
00:01:47,360 --> 00:01:51,680
Speaker 1: message that claims that this hacker gave GAB CEO Andrew

27
00:01:51,760 --> 00:01:56,400
Speaker 1: Torba and ultimatum cough up eight bitcoins, which is worth

28
00:01:56,520 --> 00:02:00,880
Speaker 1: about four fifty thousand dollars, and the hacker would return

29
00:02:00,920 --> 00:02:05,040
Speaker 1: at least some of that stolen data. Torva apparently refused, which,

30
00:02:05,080 --> 00:02:07,920
Speaker 1: for the record, is what security experts say is the

31
00:02:08,040 --> 00:02:13,000
Speaker 1: right call. Capitulating to hacker demands typically just reinforces that

32
00:02:13,080 --> 00:02:16,800
Speaker 1: hackers decisions, and it also encourages other hackers to follow suit,

33
00:02:17,360 --> 00:02:20,400
Speaker 1: because if you pay the ransom, people know that there's

34
00:02:20,400 --> 00:02:23,200
Speaker 1: money to be made, and it just makes things worse. Plus,

35
00:02:23,520 --> 00:02:26,320
Speaker 1: there's never actually a guarantee that you'll ever get anything

36
00:02:26,360 --> 00:02:30,000
Speaker 1: back anyway, or that there won't be copies of the

37
00:02:30,040 --> 00:02:32,880
Speaker 1: stuff you get back floating around now. I have to

38
00:02:32,919 --> 00:02:37,959
Speaker 1: say GAB definitely comes out looking like a terrible steward

39
00:02:38,120 --> 00:02:43,240
Speaker 1: of user security and data like absolutely awful. The company

40
00:02:43,280 --> 00:02:47,320
Speaker 1: has completely failed to protect user assets. And just to

41
00:02:47,320 --> 00:02:50,520
Speaker 1: be clear, while I am about as far away in

42
00:02:50,600 --> 00:02:56,519
Speaker 1: political ideology as you can get from the typical GAB user,

43
00:02:57,040 --> 00:03:00,840
Speaker 1: I also don't really care for hackers who promise sites

44
00:03:00,960 --> 00:03:04,680
Speaker 1: and then issue ransom notices. I don't think that's a

45
00:03:04,760 --> 00:03:08,080
Speaker 1: valid approach either. I don't think anyone comes out of

46
00:03:08,080 --> 00:03:11,280
Speaker 1: this situation looking like a good guy, and that's gonna

47
00:03:11,320 --> 00:03:14,400
Speaker 1: be a kind of common message in today's news items.

48
00:03:14,440 --> 00:03:17,519
Speaker 1: I think my guess is that the leaders at gab

49
00:03:17,560 --> 00:03:20,920
Speaker 1: are kind of in over their heads. They got the

50
00:03:20,960 --> 00:03:25,480
Speaker 1: boot from various other platforms due to the actions of

51
00:03:25,520 --> 00:03:29,239
Speaker 1: their users and different violations of terms of service, so

52
00:03:29,400 --> 00:03:32,040
Speaker 1: they're kind of forced to take whatever they can get,

53
00:03:32,520 --> 00:03:36,520
Speaker 1: which is something we've also seen with Parlor. Moving on

54
00:03:36,560 --> 00:03:39,640
Speaker 1: to a different hacking story, I've covered the Solar Winds

55
00:03:39,720 --> 00:03:43,720
Speaker 1: hack numerous times already about how it looks as though

56
00:03:44,120 --> 00:03:47,920
Speaker 1: Russian backed hackers. I mean, it's all but confirmed that

57
00:03:47,960 --> 00:03:51,520
Speaker 1: they are Russian backed hackers who compromise Solar Winds software

58
00:03:51,600 --> 00:03:55,200
Speaker 1: build system and that let the hackers push out software

59
00:03:55,320 --> 00:03:59,400
Speaker 1: updates as if they were legitimate updates. But those updates

60
00:03:59,480 --> 00:04:03,720
Speaker 1: really carried malware two Solar Winds customers, and that gave

61
00:04:03,760 --> 00:04:07,760
Speaker 1: the hackers the ability to infiltrate thousands of computer systems

62
00:04:07,840 --> 00:04:10,400
Speaker 1: and they could then follow up on that attack. This

63
00:04:10,600 --> 00:04:14,840
Speaker 1: was the supply chain attack, where targeted systems accepted the

64
00:04:14,880 --> 00:04:18,680
Speaker 1: malware because it was coming from a previously trusted source,

65
00:04:18,839 --> 00:04:23,400
Speaker 1: that being Solar Winds. Well, now cybersecurity researchers have published

66
00:04:23,400 --> 00:04:27,000
Speaker 1: a report that shows hackers backed by China we're targeting

67
00:04:27,040 --> 00:04:30,560
Speaker 1: Solar Winds at the same time in a separate operation,

68
00:04:30,920 --> 00:04:35,760
Speaker 1: one that I have previously conflated with the Russian hackers.

69
00:04:35,839 --> 00:04:38,560
Speaker 1: So it's time to set the record straight. Based on

70
00:04:38,720 --> 00:04:41,760
Speaker 1: what these researchers found. While the Russian hack used the

71
00:04:41,760 --> 00:04:44,800
Speaker 1: method I just described a moment ago, it appears that

72
00:04:44,839 --> 00:04:48,440
Speaker 1: it was the Chinese hackers who were specifically targeting a

73
00:04:48,520 --> 00:04:54,240
Speaker 1: software project called Ryan and these hackers, according to the researchers,

74
00:04:54,680 --> 00:04:59,280
Speaker 1: were really going after a specific Solar Winds customer and

75
00:04:59,320 --> 00:05:04,880
Speaker 1: they install a malware shell now called Supernova around that

76
00:05:04,960 --> 00:05:08,360
Speaker 1: Orion software. So this sounds like it was a much

77
00:05:08,360 --> 00:05:12,320
Speaker 1: more targeted attack. The Russian approach was different. It let

78
00:05:12,480 --> 00:05:18,000
Speaker 1: hackers blast out malware two thousands of potential targets, and

79
00:05:18,040 --> 00:05:21,960
Speaker 1: then the hackers could follow up with the specific hits

80
00:05:22,000 --> 00:05:24,680
Speaker 1: that they wanted to. They could look at who they

81
00:05:24,680 --> 00:05:29,200
Speaker 1: were able to capture with that first blast and follow

82
00:05:29,320 --> 00:05:34,080
Speaker 1: up for further infiltration. Whereas the Chinese approach was different.

83
00:05:34,160 --> 00:05:36,880
Speaker 1: They were looking at a specific target and went right

84
00:05:36,920 --> 00:05:41,279
Speaker 1: for them, as opposed to doing that blast attack. And

85
00:05:41,440 --> 00:05:45,000
Speaker 1: we are not done with the hacking stories just yet.

86
00:05:45,760 --> 00:05:50,279
Speaker 1: Vercada or Verkada is a Silicon Valley startup that sells

87
00:05:50,360 --> 00:05:54,800
Speaker 1: security cameras with cloud based services, and they were hacked

88
00:05:54,880 --> 00:05:58,359
Speaker 1: earlier this week. Now I say hacked, but when I

89
00:05:58,400 --> 00:06:00,520
Speaker 1: tell you how it actually happened you, I think that

90
00:06:01,000 --> 00:06:03,479
Speaker 1: hack might be a bit of a generous term for

91
00:06:04,080 --> 00:06:09,240
Speaker 1: this particular thing. For Kado practices terrible security, and for

92
00:06:09,279 --> 00:06:12,479
Speaker 1: a company that's in the security business, that's not great.

93
00:06:12,839 --> 00:06:16,640
Speaker 1: So the hackers responsible belonged to a group called APT

94
00:06:17,160 --> 00:06:22,960
Speaker 1: six nine four to zero arson cats. Okay, but APT,

95
00:06:23,120 --> 00:06:25,719
Speaker 1: by the way, stands for Advanced Persistent Threat. It's a

96
00:06:25,720 --> 00:06:30,000
Speaker 1: common term in the cybersecurity world. Anyway, these hackers discovered

97
00:06:30,080 --> 00:06:34,360
Speaker 1: that Verkada had an Internet portal for the company's internal

98
00:06:34,440 --> 00:06:38,680
Speaker 1: development system away to log in to make changes to

99
00:06:38,800 --> 00:06:42,320
Speaker 1: code and various Arcadia products. You know, it's the way

100
00:06:42,360 --> 00:06:46,719
Speaker 1: that the developers can access stuff. Now, that makes sense,

101
00:06:47,200 --> 00:06:49,560
Speaker 1: particularly in a world where presumably a lot of the

102
00:06:49,600 --> 00:06:52,800
Speaker 1: people working for Vercada are doing so remotely. But what

103
00:06:53,000 --> 00:06:56,320
Speaker 1: does not make sense is that the hackers say that

104
00:06:56,360 --> 00:07:00,520
Speaker 1: this was essentially a publicly accessible portal. There were no

105
00:07:00,680 --> 00:07:05,760
Speaker 1: log in credentials required to get to that portal system,

106
00:07:05,800 --> 00:07:09,560
Speaker 1: so you get there without having to first verify that

107
00:07:09,640 --> 00:07:13,640
Speaker 1: you work for the company, and on that landing page

108
00:07:13,720 --> 00:07:16,840
Speaker 1: or that landing site, they found the login credentials to

109
00:07:16,880 --> 00:07:21,840
Speaker 1: get what they called super admin level access to Arcada's systems.

110
00:07:21,840 --> 00:07:24,600
Speaker 1: So as a result, they were able to access more

111
00:07:24,720 --> 00:07:28,840
Speaker 1: than one hundred thousand cameras the number I frequently see

112
00:07:28,840 --> 00:07:33,400
Speaker 1: a one hundred fifty thousand camera feeds belonging to various

113
00:07:33,480 --> 00:07:37,120
Speaker 1: Arcaded customers. They also got a list that was around

114
00:07:37,120 --> 00:07:41,280
Speaker 1: twenty four thousand entries long that that names those customers,

115
00:07:41,360 --> 00:07:47,120
Speaker 1: and they include businesses, churches, health care facilities, jails, and

116
00:07:47,280 --> 00:07:51,040
Speaker 1: the p g A of all things. Tesla is one

117
00:07:51,080 --> 00:07:54,720
Speaker 1: of their customers. Though Tesla says that the hack really

118
00:07:54,760 --> 00:07:57,720
Speaker 1: just gave the hackers of you into one of Tesla's

119
00:07:57,880 --> 00:08:03,320
Speaker 1: suppliers sites, but not the main manufacturing facility in Shanghai.

120
00:08:03,720 --> 00:08:08,080
Speaker 1: A software developer named Tilly Kottman gave details about the hack,

121
00:08:08,480 --> 00:08:12,320
Speaker 1: presumably having played some part in carrying it out. Now,

122
00:08:12,320 --> 00:08:14,280
Speaker 1: what we call this a hack? I would argue that

123
00:08:14,360 --> 00:08:18,480
Speaker 1: finding log in credentials on a publicly accessible landing page

124
00:08:19,040 --> 00:08:22,680
Speaker 1: is pretty much like walking up to someone's password protected

125
00:08:22,720 --> 00:08:26,240
Speaker 1: computer and seeing that they put the log in and

126
00:08:26,320 --> 00:08:29,640
Speaker 1: password on a sticky note and stuck it to the monitor.

127
00:08:30,120 --> 00:08:32,960
Speaker 1: It's not exactly safe. I mean, there's no point in

128
00:08:33,080 --> 00:08:37,560
Speaker 1: having a locked door if you're hanging a key off

129
00:08:37,600 --> 00:08:41,160
Speaker 1: the door. Knob Kottman says that part of the reason

130
00:08:41,280 --> 00:08:44,760
Speaker 1: that the hacker group published the information and shared video

131
00:08:44,880 --> 00:08:48,800
Speaker 1: from various locations you know Varcada customers, is that they

132
00:08:48,880 --> 00:08:52,920
Speaker 1: wanted to point out how widely distributed Varcada's systems actually

133
00:08:52,960 --> 00:08:57,079
Speaker 1: are and how inherently unsafe they are. Now, it's hard

134
00:08:57,160 --> 00:09:00,000
Speaker 1: for me to disagree with those points. I would argue

135
00:09:00,000 --> 00:09:02,360
Speaker 1: you this kind of puts the hackers into a sort

136
00:09:02,400 --> 00:09:06,360
Speaker 1: of gray hat area when it comes to the spectrum

137
00:09:06,480 --> 00:09:10,320
Speaker 1: of hackers, so we typically describe them by hats. So

138
00:09:10,360 --> 00:09:14,000
Speaker 1: you've got white hat hackers. These are hackers who probit systems.

139
00:09:14,040 --> 00:09:18,000
Speaker 1: They look for vulnerabilities, but the intent is to tell

140
00:09:18,280 --> 00:09:23,559
Speaker 1: the respective system administrators about those vulnerabilities, hopefully before someone

141
00:09:23,600 --> 00:09:26,800
Speaker 1: else can exploit them. So the whole goal is to

142
00:09:26,840 --> 00:09:29,360
Speaker 1: find weaknesses and systems and then say, hey, you need

143
00:09:29,400 --> 00:09:32,680
Speaker 1: to fix this. But then you've got your black hat hackers.

144
00:09:32,679 --> 00:09:35,160
Speaker 1: These are the people who are trying to profit from

145
00:09:35,160 --> 00:09:38,600
Speaker 1: being able to access systems and exploit them in some way.

146
00:09:39,200 --> 00:09:42,560
Speaker 1: Gray hats are somewhere in the middle. The hackers didn't

147
00:09:42,720 --> 00:09:45,880
Speaker 1: just go to Rakeda to alert the company of the mistake.

148
00:09:46,320 --> 00:09:49,240
Speaker 1: They didn't go to say, hey, you've got this massive

149
00:09:49,280 --> 00:09:52,400
Speaker 1: security vulnerability, you need to fix it right now. They

150
00:09:52,440 --> 00:09:56,720
Speaker 1: went public with this revelation, which definitely makes the company

151
00:09:56,760 --> 00:09:59,319
Speaker 1: look bad, And honestly, I think you can make a

152
00:09:59,360 --> 00:10:02,720
Speaker 1: good argument that there's some merit in that, considering that

153
00:10:02,760 --> 00:10:07,080
Speaker 1: the whole value proposition for a security company is that

154
00:10:07,600 --> 00:10:11,360
Speaker 1: it's safe. This is also a good reminder that security

155
00:10:11,360 --> 00:10:15,400
Speaker 1: systems that include ways to access a camera feed remotely

156
00:10:15,960 --> 00:10:19,920
Speaker 1: represents a potential security vulnerability. It's always a good idea

157
00:10:20,400 --> 00:10:23,839
Speaker 1: to do your research before you choose a security solution

158
00:10:24,440 --> 00:10:27,880
Speaker 1: on a related note, Jason Cobler and Joseph Cox over

159
00:10:27,920 --> 00:10:30,680
Speaker 1: at Vice dot Com pointed out that another really big

160
00:10:30,679 --> 00:10:34,760
Speaker 1: issue with this hack is that Verkada offers facial recognition

161
00:10:34,920 --> 00:10:39,280
Speaker 1: solutions in their security camera technologies, and that means that

162
00:10:39,360 --> 00:10:42,000
Speaker 1: hackers weren't just able to look at video feeds, they

163
00:10:42,040 --> 00:10:45,560
Speaker 1: could potentially identify the people who showed up in those

164
00:10:45,640 --> 00:10:49,040
Speaker 1: video feeds, And as they pointed out, quote, the breach

165
00:10:49,200 --> 00:10:53,040
Speaker 1: shows the astonishing reach of facial recognition enabled cameras in

166
00:10:53,160 --> 00:10:59,000
Speaker 1: ordinary workplaces, bars, parking lots, schools, stores, and more. End quote.

167
00:10:59,520 --> 00:11:02,640
Speaker 1: I think that that's putting it lightly because honestly, while

168
00:11:03,000 --> 00:11:06,240
Speaker 1: this is all about security cameras and a company that

169
00:11:06,720 --> 00:11:10,440
Speaker 1: has this kind of proprietary approach to facial recognition, we

170
00:11:10,480 --> 00:11:13,280
Speaker 1: have to remember that just about everybody carries a camera

171
00:11:13,400 --> 00:11:17,320
Speaker 1: with them, and depending upon the apps being used, a

172
00:11:17,360 --> 00:11:20,440
Speaker 1: lot of these companies are able to take advantage of

173
00:11:20,600 --> 00:11:25,040
Speaker 1: massive amounts of data and do facial recognition on their own.

174
00:11:25,360 --> 00:11:29,040
Speaker 1: So yeah, this is a very acute UH case that

175
00:11:29,120 --> 00:11:32,479
Speaker 1: we can point to, but it's by no means an outlier.

176
00:11:32,800 --> 00:11:35,760
Speaker 1: It is easy to imagine a scenario in which malicious

177
00:11:35,760 --> 00:11:39,400
Speaker 1: hackers would not only breach a system like Verkada, but

178
00:11:39,480 --> 00:11:42,880
Speaker 1: also keep it quiet, right, They might never come forward

179
00:11:43,120 --> 00:11:45,880
Speaker 1: letting people know that they got access, and then in

180
00:11:45,920 --> 00:11:48,480
Speaker 1: the meantime they could use these surveillance cameras for their

181
00:11:48,480 --> 00:11:52,800
Speaker 1: own purposes and perhaps even spy and potentially blackmail specific people.

182
00:11:53,360 --> 00:11:57,880
Speaker 1: The facial recognition genie is out of the bottle, and

183
00:11:57,920 --> 00:12:00,440
Speaker 1: the fact that there are numerous big companies are making

184
00:12:00,559 --> 00:12:03,880
Speaker 1: use of the technology in a widely distributed way means

185
00:12:03,920 --> 00:12:08,040
Speaker 1: that whenever you're on camera, you are potentially identifiable in

186
00:12:08,120 --> 00:12:11,880
Speaker 1: real time, and you're pretty much always just moments away

187
00:12:11,880 --> 00:12:14,960
Speaker 1: from being on camera if you're out and about. So

188
00:12:15,080 --> 00:12:18,320
Speaker 1: fun times. So hey, let's stay on this topic for

189
00:12:18,360 --> 00:12:21,760
Speaker 1: a little bit. The United States Army Research Laboratory has

190
00:12:21,760 --> 00:12:25,240
Speaker 1: been working on image recognition AI applications that will be

191
00:12:25,280 --> 00:12:28,360
Speaker 1: able to identify faces even if those images were taken

192
00:12:28,960 --> 00:12:33,360
Speaker 1: in darkness. So this research team took half a million

193
00:12:33,400 --> 00:12:37,120
Speaker 1: pictures of three people, which is a pretty small sample size,

194
00:12:37,120 --> 00:12:39,439
Speaker 1: believe it or not. Now, some of those photos were

195
00:12:39,480 --> 00:12:43,160
Speaker 1: taken in normal conditions, normal lighting conditions using a standard camera.

196
00:12:43,559 --> 00:12:46,200
Speaker 1: Others were taken in low light conditions, and some with

197
00:12:46,320 --> 00:12:49,960
Speaker 1: thermal cameras. I think it's pretty obvious to see where

198
00:12:49,960 --> 00:12:53,240
Speaker 1: the benefits are from a military perspective of having technology

199
00:12:53,280 --> 00:12:56,760
Speaker 1: that can identify a person even in low lighting conditions,

200
00:12:57,120 --> 00:13:00,120
Speaker 1: But there is no denying the idea is more or

201
00:13:00,120 --> 00:13:03,480
Speaker 1: than a little creepy. That being said, half a million photos,

202
00:13:03,520 --> 00:13:05,800
Speaker 1: like I said, is a very small sample size, and

203
00:13:05,840 --> 00:13:08,800
Speaker 1: the team says they're making progress, but they are nowhere

204
00:13:08,800 --> 00:13:12,080
Speaker 1: near a level where anything is sophisticated enough for deployment.

205
00:13:12,440 --> 00:13:15,160
Speaker 1: The system is still struggling to identify images and low

206
00:13:15,240 --> 00:13:18,840
Speaker 1: lighting conditions. Thermal cameras produce very different kinds of photos

207
00:13:18,880 --> 00:13:21,840
Speaker 1: than our normal cameras do, and the computer systems haven't

208
00:13:21,880 --> 00:13:25,360
Speaker 1: quite figured out how to reliably map those thermal images

209
00:13:25,640 --> 00:13:29,040
Speaker 1: to specific people. Also, they said that just a small

210
00:13:29,160 --> 00:13:32,000
Speaker 1: change in the camera's viewing angle, like the angle between

211
00:13:32,120 --> 00:13:34,280
Speaker 1: the person's face and the camera, can make it a

212
00:13:34,320 --> 00:13:37,079
Speaker 1: lot harder for a computer to suss out who they

213
00:13:37,120 --> 00:13:41,440
Speaker 1: are looking at. I've often talked about image recognition by

214
00:13:41,520 --> 00:13:44,080
Speaker 1: using coffee mugs as kind of an example. If you

215
00:13:44,120 --> 00:13:47,240
Speaker 1: were to feed millions of images of red coffee mugs

216
00:13:47,240 --> 00:13:49,559
Speaker 1: to a computer, but every single one of those images

217
00:13:50,000 --> 00:13:53,280
Speaker 1: showed the red coffee mug having it's handled pointing off

218
00:13:53,280 --> 00:13:56,160
Speaker 1: to the right side in some way, then the computer

219
00:13:56,280 --> 00:13:59,599
Speaker 1: might balk at seeing that same red coffee mug. But

220
00:13:59,679 --> 00:14:02,240
Speaker 1: what they handle pointed to the left side that could

221
00:14:02,240 --> 00:14:05,079
Speaker 1: be enough to throw the computer off. Computers are remarkable,

222
00:14:05,360 --> 00:14:08,400
Speaker 1: but they can still be pretty dumb in some ways. Still,

223
00:14:08,440 --> 00:14:11,559
Speaker 1: this area of research is a bit scary, particularly since

224
00:14:11,840 --> 00:14:14,120
Speaker 1: we already live in a world where lots of entities

225
00:14:14,880 --> 00:14:19,240
Speaker 1: like law enforcement agencies, rely on facial recognition technologies. And

226
00:14:19,280 --> 00:14:22,160
Speaker 1: that's without even getting into the existing problems we already

227
00:14:22,160 --> 00:14:25,720
Speaker 1: have with facial recognition, like racial and gender bias that

228
00:14:25,720 --> 00:14:29,720
Speaker 1: can lead to inaccurate results. And gosh, I wish I

229
00:14:29,760 --> 00:14:32,560
Speaker 1: had a happier story I could segue too. But I

230
00:14:32,560 --> 00:14:35,560
Speaker 1: should also mention that the Los Angeles Times published an

231
00:14:35,600 --> 00:14:39,520
Speaker 1: article titled clear View AI uses your online photos to

232
00:14:39,640 --> 00:14:44,440
Speaker 1: instantly I du that's a problem, lawsuit says, And yeah,

233
00:14:44,520 --> 00:14:46,960
Speaker 1: the headline pretty much tells the story. So you've got

234
00:14:46,960 --> 00:14:50,840
Speaker 1: this company, clear View AI, and they have an an

235
00:14:50,960 --> 00:14:55,240
Speaker 1: enormous image database, and they created it by scraping various

236
00:14:56,000 --> 00:15:00,240
Speaker 1: websites and services, particularly social networking platforms like face Book

237
00:15:00,240 --> 00:15:04,240
Speaker 1: and Twitter, but also other types of services like Google,

238
00:15:04,320 --> 00:15:07,960
Speaker 1: and Venmo, and they started gathering photos that way. According

239
00:15:07,960 --> 00:15:10,240
Speaker 1: to the l A Times, that means that the company

240
00:15:10,320 --> 00:15:13,880
Speaker 1: has a database of more than three billion photos and

241
00:15:13,920 --> 00:15:17,200
Speaker 1: has software that creates a digital faceprint of each person

242
00:15:17,320 --> 00:15:21,200
Speaker 1: based on those photos, which allows for faster facial recognition

243
00:15:21,240 --> 00:15:25,480
Speaker 1: identification if the system encounters a new image of someone

244
00:15:25,520 --> 00:15:28,920
Speaker 1: who is already in the database somewhere. So all those

245
00:15:28,920 --> 00:15:31,760
Speaker 1: photos that people share on different sites without really thinking

246
00:15:31,760 --> 00:15:34,400
Speaker 1: about it, and I include myself in this group of people,

247
00:15:35,120 --> 00:15:38,200
Speaker 1: you could be part of that massive database. That means

248
00:15:38,240 --> 00:15:41,400
Speaker 1: this company could potentially be using those photos to make

249
00:15:41,440 --> 00:15:44,800
Speaker 1: it possible for a clear view customers like once again,

250
00:15:45,040 --> 00:15:49,040
Speaker 1: law enforcement agencies to use it in real time with

251
00:15:49,160 --> 00:15:53,120
Speaker 1: the various solutions. Now, this prompted some civil liberties activists

252
00:15:53,160 --> 00:15:56,640
Speaker 1: to file a lawsuit against the company in California. They

253
00:15:56,680 --> 00:16:00,480
Speaker 1: said that the company's practices violate privacy and create a

254
00:16:00,560 --> 00:16:05,080
Speaker 1: chilling effect on protected activities such as the right to assembly. Now,

255
00:16:05,160 --> 00:16:07,360
Speaker 1: right now, we're in a pandemic. People shouldn't really be

256
00:16:07,400 --> 00:16:10,960
Speaker 1: assembling in public and big numbers anyway. But they're specifically

257
00:16:11,000 --> 00:16:14,760
Speaker 1: talking about things like the Black Lives Matter movement. So

258
00:16:15,000 --> 00:16:19,280
Speaker 1: the people who have had their accounts scraped, were never

259
00:16:19,320 --> 00:16:22,520
Speaker 1: given any notice that that was happening, let alone ever

260
00:16:22,640 --> 00:16:26,800
Speaker 1: given the chance to give consent for it. Now, the

261
00:16:26,840 --> 00:16:29,360
Speaker 1: flip side of that argument is that you could say,

262
00:16:29,520 --> 00:16:32,480
Speaker 1: someone posting to Facebook or whatever is doing so in

263
00:16:32,520 --> 00:16:35,920
Speaker 1: a semi public way, right if they haven't protected their

264
00:16:35,920 --> 00:16:40,000
Speaker 1: account in any way, then anyone could potentially see those photos.

265
00:16:40,040 --> 00:16:42,000
Speaker 1: But then you have to ask, what about people who

266
00:16:42,040 --> 00:16:45,480
Speaker 1: don't have a Facebook account. Maybe their friends have taken

267
00:16:45,520 --> 00:16:47,880
Speaker 1: photos of them and put those photos up on their

268
00:16:47,920 --> 00:16:52,040
Speaker 1: own account. Maybe those photos even have identifiable information in them,

269
00:16:52,080 --> 00:16:54,040
Speaker 1: And then you've got someone who doesn't even have a

270
00:16:54,080 --> 00:16:57,520
Speaker 1: Facebook account who had their picture scraped for this kind

271
00:16:57,520 --> 00:17:01,520
Speaker 1: of thing doesn't matter in that case. So the plaintiffs

272
00:17:01,560 --> 00:17:05,480
Speaker 1: are focusing just on California and California's citizens for this

273
00:17:05,520 --> 00:17:09,439
Speaker 1: particular lawsuit, and they are seeking an injunction that would

274
00:17:09,520 --> 00:17:13,960
Speaker 1: force clear View AI to stop collecting biometric information on

275
00:17:14,200 --> 00:17:17,159
Speaker 1: people in California, But they also want the company to

276
00:17:17,240 --> 00:17:20,760
Speaker 1: delete all the biometric data that it has already collected.

277
00:17:21,200 --> 00:17:25,200
Speaker 1: Clear View AI faces a similar lawsuit in Illinois. That

278
00:17:25,240 --> 00:17:27,720
Speaker 1: one is being brought against it by the American Civil

279
00:17:27,800 --> 00:17:30,679
Speaker 1: Liberties Union. I'm sure. I will be following up on

280
00:17:30,720 --> 00:17:33,720
Speaker 1: this story later on in the year as it plays out.

281
00:17:34,320 --> 00:17:39,800
Speaker 1: It really is indicative of how people are becoming aware

282
00:17:40,400 --> 00:17:44,560
Speaker 1: that the early choices we made when we were building

283
00:17:44,560 --> 00:17:51,160
Speaker 1: out stuff like social networking sites, we're rather shortsighted and

284
00:17:51,240 --> 00:17:55,720
Speaker 1: have had bigger consequences than we anticipated back then. I mean, heck,

285
00:17:55,760 --> 00:18:00,200
Speaker 1: Facebook started as a website for students to rate how

286
00:18:00,240 --> 00:18:04,560
Speaker 1: hot each other happened to be, So that's kind of

287
00:18:04,560 --> 00:18:07,120
Speaker 1: creepy all on its own. It doesn't have the same

288
00:18:07,160 --> 00:18:09,720
Speaker 1: sort of weight as this is a site that some

289
00:18:09,800 --> 00:18:12,439
Speaker 1: company is going to use later down the road to

290
00:18:12,600 --> 00:18:17,800
Speaker 1: build out an identification engine that law enforcement agencies might

291
00:18:17,840 --> 00:18:22,199
Speaker 1: be using ethically or otherwise in the future. Uh. And

292
00:18:22,280 --> 00:18:25,120
Speaker 1: because of those consequences, we now have to revisit those

293
00:18:25,119 --> 00:18:27,840
Speaker 1: early decisions we made and ask questions like, how can

294
00:18:27,880 --> 00:18:30,800
Speaker 1: we address this? Are there any ways to fix it?

295
00:18:31,200 --> 00:18:34,160
Speaker 1: Will it require us to create a ban on the

296
00:18:34,200 --> 00:18:40,120
Speaker 1: private use of facial recognition technologies? Honestly, those are big

297
00:18:40,200 --> 00:18:42,280
Speaker 1: questions that we don't have answers to yet, and I

298
00:18:42,320 --> 00:18:45,119
Speaker 1: don't know what answers we will arrive at, but I

299
00:18:45,160 --> 00:18:49,720
Speaker 1: will continue to cover the cases. That's it for today's news.

300
00:18:49,760 --> 00:18:52,919
Speaker 1: I hope you guys stay safe and healthy. If you

301
00:18:52,960 --> 00:18:55,040
Speaker 1: have anything you want to share with me, the best

302
00:18:55,040 --> 00:18:57,760
Speaker 1: place to do it is over on Twitter the handle

303
00:18:57,800 --> 00:19:00,639
Speaker 1: I use this tech stuff. H S double you and

304
00:19:00,680 --> 00:19:08,560
Speaker 1: I'll talk to you again really soon. Y Text Stuff

305
00:19:08,640 --> 00:19:11,800
Speaker 1: is an I Heart Radio production. For more podcasts from

306
00:19:11,800 --> 00:19:15,600
Speaker 1: my Heart Radio, visit the I Heart Radio app, Apple Podcasts,

307
00:19:15,680 --> 00:19:17,679
Speaker 1: or wherever you listen to your favorite shows