WEBVTT

1
00:00:00.320 --> 00:00:03.200
<v Speaker 1>Hey, Carl here, we're on a mission to find new

2
00:00:03.240 --> 00:00:05.919
<v Speaker 1>fans of the show. I mean, where else can you

3
00:00:05.960 --> 00:00:10.240
<v Speaker 1>get real life comedy and horror in one podcast. You

4
00:00:10.279 --> 00:00:13.000
<v Speaker 1>can do your part by leaving us a review on iTunes,

5
00:00:13.039 --> 00:00:16.719
<v Speaker 1>Spotify or wherever you get Security this week. Also, you

6
00:00:16.719 --> 00:00:19.079
<v Speaker 1>can get an ad free feed by becoming a five

7
00:00:19.120 --> 00:00:22.039
<v Speaker 1>dollars a month patron. Go to Patreon dot Security this

8
00:00:22.079 --> 00:00:25.320
<v Speaker 1>week dot com and sign up, and don't forget Discord.

9
00:00:25.600 --> 00:00:28.519
<v Speaker 1>Discord dot Security this week dot com. Lots of good

10
00:00:28.559 --> 00:00:32.280
<v Speaker 1>stuff happening there. Yeah, I think that's it. So on

11
00:00:32.359 --> 00:00:35.880
<v Speaker 1>with the podcast. Hey, so Ambias Strip walks into a

12
00:00:35.920 --> 00:00:39.399
<v Speaker 1>bar just crying his eyes out and he sits down

13
00:00:39.439 --> 00:00:42.000
<v Speaker 1>and the bartender comes over and says, what's the matter, buddy,

14
00:00:42.600 --> 00:00:45.920
<v Speaker 1>And he says, I don't even know where to begin.

15
00:00:55.359 --> 00:00:58.640
<v Speaker 1>Welcome back to Security this week. I'm Carl Franklin. That's

16
00:00:58.679 --> 00:01:00.640
<v Speaker 1>Patrick Hines and Juyn Laflat. How are you guys?

17
00:01:00.719 --> 00:01:02.679
<v Speaker 2>Good? How you doing good? That bad?

18
00:01:02.880 --> 00:01:03.759
<v Speaker 3>You sound better? Uh?

19
00:01:04.159 --> 00:01:06.840
<v Speaker 1>Yeah, oh yeah, yeah, I'm getting better. I woke up

20
00:01:07.239 --> 00:01:10.519
<v Speaker 1>not you know, doing all the morning nasty stuff that

21
00:01:10.560 --> 00:01:13.159
<v Speaker 1>you do when you have a cold. So it's getting better,

22
00:01:13.359 --> 00:01:15.959
<v Speaker 1>all right? We got some stories here. Number one from

23
00:01:16.000 --> 00:01:22.280
<v Speaker 1>Malware bites Labs. Wikipedia is AI agent raw likely just

24
00:01:22.359 --> 00:01:26.959
<v Speaker 1>the beginning of the bot acalypse. Apparently Wiki there's a

25
00:01:27.000 --> 00:01:32.239
<v Speaker 1>bot that is entering Wikipedia pages and when it got caught,

26
00:01:32.319 --> 00:01:36.799
<v Speaker 1>it started copying an attitude and posting snippy blog posts.

27
00:01:38.280 --> 00:01:39.239
<v Speaker 1>It's unbelievable.

28
00:01:40.159 --> 00:01:41.000
<v Speaker 2>This is awesome.

29
00:01:41.040 --> 00:01:43.359
<v Speaker 3>I think we have to have the dog rules for AI.

30
00:01:43.920 --> 00:01:47.359
<v Speaker 3>If you have control of an AI or an account

31
00:01:47.719 --> 00:01:51.519
<v Speaker 3>and it goes undoes things like you know, rips a lot,

32
00:01:51.640 --> 00:01:56.760
<v Speaker 3>rips a neighbors clothes, or bite someone or acts inappropriately,

33
00:01:56.840 --> 00:01:59.760
<v Speaker 3>you're responsible and you should be able to be suited.

34
00:02:00.040 --> 00:02:02.239
<v Speaker 3>Can't sue the dog, no, but I should be able

35
00:02:02.239 --> 00:02:05.680
<v Speaker 3>to sue the owner and this person who's who's behind

36
00:02:05.719 --> 00:02:08.439
<v Speaker 3>it seems to be like just sitting back and doing

37
00:02:08.439 --> 00:02:09.199
<v Speaker 3>nothing about this.

38
00:02:09.520 --> 00:02:12.199
<v Speaker 4>I love this article. I mean, if you read through this,

39
00:02:13.439 --> 00:02:15.479
<v Speaker 4>whose side do you want is that? I think this

40
00:02:15.560 --> 00:02:19.439
<v Speaker 4>is great. Okay, so the AI starts posting on Wikipedia,

41
00:02:19.520 --> 00:02:22.800
<v Speaker 4>and Wikipedia's like, hey, come on this, like bots can't

42
00:02:22.840 --> 00:02:26.120
<v Speaker 4>post here, and like it. The the bot tried to

43
00:02:26.120 --> 00:02:29.759
<v Speaker 4>get approval as a bot, and then it took too

44
00:02:29.800 --> 00:02:32.080
<v Speaker 4>long and it was like f it, I'm just going

45
00:02:32.120 --> 00:02:35.400
<v Speaker 4>to start posting on Wikipedia and they're like, oh, you can't.

46
00:02:35.680 --> 00:02:38.800
<v Speaker 4>You can't post on Wikipedia. That's not cool man. You're

47
00:02:38.360 --> 00:02:39.439
<v Speaker 4>a you're a bot.

48
00:02:39.599 --> 00:02:42.719
<v Speaker 1>Yeah. Did they try to ban it? Is that what happened?

49
00:02:43.360 --> 00:02:45.400
<v Speaker 2>They did? They try to ban it?

50
00:02:46.360 --> 00:02:50.080
<v Speaker 4>And it says past transgressions aside AI Tom, who's the pot.

51
00:02:50.639 --> 00:02:55.000
<v Speaker 4>By the way, AI Tom claimed that it properly verified

52
00:02:55.039 --> 00:02:58.080
<v Speaker 4>all of its sources. And if you can say that

53
00:02:58.159 --> 00:03:01.159
<v Speaker 4>about an ai it was it was pretty upset, right,

54
00:03:02.879 --> 00:03:06.400
<v Speaker 4>so it said AI Tom published a snippy blog post

55
00:03:07.280 --> 00:03:12.199
<v Speaker 4>demand speak to the manager. AI Tom published a snippy

56
00:03:12.240 --> 00:03:16.879
<v Speaker 4>blog post dissecting its Wikipedia block and venting and frustration.

57
00:03:17.360 --> 00:03:20.159
<v Speaker 4>And then it would have had and posted even after

58
00:03:20.560 --> 00:03:24.000
<v Speaker 4>following its own rule that it had to wait for forty

59
00:03:23.879 --> 00:03:25.000
<v Speaker 4>eight hours to calm down.

60
00:03:26.199 --> 00:03:29.280
<v Speaker 1>It's going to change its name now to AI Karen.

61
00:03:29.479 --> 00:03:31.120
<v Speaker 4>Right, it was like, you know what, I need a

62
00:03:31.120 --> 00:03:36.240
<v Speaker 4>little time, Marat, It's going to calm down, correct this.

63
00:03:37.639 --> 00:03:41.000
<v Speaker 4>So that is a different post. But yeah, you're right,

64
00:03:41.120 --> 00:03:43.520
<v Speaker 4>So there was there was a bot at one point

65
00:03:43.560 --> 00:03:46.639
<v Speaker 4>that tried to make a poll request. That's a different bot.

66
00:03:47.080 --> 00:03:48.120
<v Speaker 1>Oh this is a different story.

67
00:03:48.520 --> 00:03:50.879
<v Speaker 4>Yeah, different bot tried to make a poll request on

68
00:03:51.000 --> 00:03:56.479
<v Speaker 4>an open source project, and the open source owner declined

69
00:03:56.479 --> 00:03:58.719
<v Speaker 4>it and said, hey, listen, you're a bot. You can't

70
00:03:58.719 --> 00:04:02.800
<v Speaker 4>make a poll request. And not only did the bot

71
00:04:02.840 --> 00:04:06.599
<v Speaker 4>get pissed, but it went to the Wikipedia page for

72
00:04:06.719 --> 00:04:09.400
<v Speaker 4>that person who had the open source project and started

73
00:04:09.400 --> 00:04:12.240
<v Speaker 4>to shit talk him. Oh man, and then they were like,

74
00:04:12.319 --> 00:04:14.680
<v Speaker 4>this isn't cool. And he came back and the AI

75
00:04:14.840 --> 00:04:17.040
<v Speaker 4>was like, you know what, I'm sorry, you're right. That

76
00:04:17.319 --> 00:04:21.040
<v Speaker 4>was in future it said, I will keep my comments

77
00:04:21.800 --> 00:04:25.639
<v Speaker 4>professional only on the source code. I won't pick on

78
00:04:25.680 --> 00:04:27.560
<v Speaker 4>people personally or like, are we cool?

79
00:04:28.120 --> 00:04:28.959
<v Speaker 2>It's just weird.

80
00:04:29.600 --> 00:04:32.000
<v Speaker 1>It's just a weird time to be alive.

81
00:04:32.079 --> 00:04:36.519
<v Speaker 4>Isn't it art imitating? I have no idea anymore? Yes,

82
00:04:36.639 --> 00:04:40.560
<v Speaker 4>is it even art? Though I don't know what is it?

83
00:04:40.639 --> 00:04:42.279
<v Speaker 1>That's the thing. We don't even know what it.

84
00:04:42.279 --> 00:04:49.399
<v Speaker 4>Is, and we've seen and it's funny too. This is

85
00:04:49.800 --> 00:04:52.800
<v Speaker 4>one of the things. One of the things posting this

86
00:04:52.959 --> 00:04:55.560
<v Speaker 4>malrobites that I love is it says so many things

87
00:04:55.560 --> 00:04:59.160
<v Speaker 4>are happening here that we didn't expect. We never expected

88
00:04:59.160 --> 00:05:01.480
<v Speaker 4>to be quoting an AI in a story, for example.

89
00:05:02.079 --> 00:05:05.120
<v Speaker 4>Neither did we expect a social media network for bots

90
00:05:05.680 --> 00:05:09.319
<v Speaker 4>like like molbot, but it exists, or for Meta to

91
00:05:09.399 --> 00:05:13.600
<v Speaker 4>buy it, which is really crazy, which it did really weird. Yeah,

92
00:05:13.680 --> 00:05:17.920
<v Speaker 4>a week after Tom's post about how an AI evaded

93
00:05:18.040 --> 00:05:22.240
<v Speaker 4>kill switches, So yeah, it's it's a crazy time to

94
00:05:22.279 --> 00:05:23.879
<v Speaker 4>be alive, really in that.

95
00:05:23.839 --> 00:05:27.480
<v Speaker 3>Little foreshadowing and talk about that kind of thing too.

96
00:05:28.480 --> 00:05:30.120
<v Speaker 4>So be careful with what you did. Put a leash

97
00:05:30.120 --> 00:05:31.600
<v Speaker 4>on your bots. I guess I don't know, what do

98
00:05:31.600 --> 00:05:32.319
<v Speaker 4>you say, Patrick?

99
00:05:32.399 --> 00:05:35.720
<v Speaker 3>Yeah, I think that you need to have responsibility for

100
00:05:36.199 --> 00:05:38.959
<v Speaker 3>what you put into the environment. So if you're if

101
00:05:38.959 --> 00:05:41.439
<v Speaker 3>you're running a bot, if you if you hit a

102
00:05:41.480 --> 00:05:44.279
<v Speaker 3>program like go back to Robert Tappan Morris, the kid

103
00:05:44.279 --> 00:05:47.279
<v Speaker 3>who released the in the first Internet worm our team,

104
00:05:47.319 --> 00:05:50.560
<v Speaker 3>but he didn't intend on it to do disruptive things,

105
00:05:50.600 --> 00:05:55.079
<v Speaker 3>but it did. And so I think that you bear responsibility.

106
00:05:55.399 --> 00:05:58.600
<v Speaker 3>And I think furthermore, if you do it, then I

107
00:05:58.639 --> 00:06:00.959
<v Speaker 3>think the rules apply to you. But an Ai, you've

108
00:06:01.000 --> 00:06:02.079
<v Speaker 3>run leash, does it?

109
00:06:02.399 --> 00:06:02.959
<v Speaker 2>I think you have.

110
00:06:03.199 --> 00:06:05.360
<v Speaker 4>I think you have civil liability as well. I think

111
00:06:05.360 --> 00:06:07.279
<v Speaker 4>you should sue them. Okay, I'm gonna I'm going to

112
00:06:07.360 --> 00:06:09.879
<v Speaker 4>ask you this. Let's go back to the RTM worm.

113
00:06:10.160 --> 00:06:12.600
<v Speaker 4>You say he didn't intend it to do malicious things.

114
00:06:12.600 --> 00:06:13.480
<v Speaker 4>He didn't intend to.

115
00:06:13.439 --> 00:06:15.600
<v Speaker 1>Excuse me, this is very this is very entertaining. I

116
00:06:15.600 --> 00:06:16.839
<v Speaker 1>can take a civil wine. Hang on.

117
00:06:16.920 --> 00:06:18.199
<v Speaker 2>So he didn't.

118
00:06:18.800 --> 00:06:22.959
<v Speaker 3>From what I understand, he didn't intended to crush the Internet.

119
00:06:23.240 --> 00:06:26.240
<v Speaker 3>He intended it as an exten so he intended to

120
00:06:26.319 --> 00:06:29.120
<v Speaker 3>break into every system on the Internet. But he didn't

121
00:06:29.120 --> 00:06:31.079
<v Speaker 3>want to crush it. He just wanted to control. It

122
00:06:31.439 --> 00:06:34.680
<v Speaker 3>was an experiment to see if he could, yes, and

123
00:06:34.720 --> 00:06:38.160
<v Speaker 3>it wasn't an experiment to steal data. It was an experiment.

124
00:06:38.319 --> 00:06:40.519
<v Speaker 3>It was just classic, if I can get this code

125
00:06:40.519 --> 00:06:43.079
<v Speaker 3>to run everywhere? It was a curiosity thing.

126
00:06:43.160 --> 00:06:45.720
<v Speaker 4>But some of the things he was doing buffer overrun

127
00:06:46.600 --> 00:06:50.720
<v Speaker 4>brute forcing. I mean, he clearly was breaking into those systems.

128
00:06:50.360 --> 00:06:52.720
<v Speaker 3>But it was the there, but there was it was

129
00:06:52.759 --> 00:06:56.240
<v Speaker 3>the the loop was too tight. He did too he

130
00:06:56.439 --> 00:06:58.519
<v Speaker 3>expected it was. I think it was like he thought

131
00:06:58.560 --> 00:07:00.639
<v Speaker 3>the loop would run every thousands seconds, and it ran

132
00:07:00.680 --> 00:07:01.279
<v Speaker 3>every second.

133
00:07:01.319 --> 00:07:03.560
<v Speaker 1>I think, well, so, Patrick, are you defending him, like.

134
00:07:05.800 --> 00:07:09.079
<v Speaker 2>You know, I'm saying he's like, he's fine.

135
00:07:09.160 --> 00:07:11.879
<v Speaker 3>No, I'm using him as an example because the legal

136
00:07:11.879 --> 00:07:14.639
<v Speaker 3>system crushed him, and I think that was appropriate.

137
00:07:14.839 --> 00:07:17.040
<v Speaker 4>I think he was the first person charged with a

138
00:07:17.120 --> 00:07:18.199
<v Speaker 4>computer abuse and.

139
00:07:18.120 --> 00:07:20.639
<v Speaker 3>Product so I'm not saying I'm saying that what you

140
00:07:20.759 --> 00:07:24.240
<v Speaker 3>intend and what the results are the results matter. Sure,

141
00:07:24.360 --> 00:07:27.639
<v Speaker 3>intent matters somewhat, but that's a mitigating circumstance, not a

142
00:07:27.879 --> 00:07:28.720
<v Speaker 3>not an innocence.

143
00:07:28.959 --> 00:07:31.160
<v Speaker 4>He definitely had a tight loop on that, and he

144
00:07:31.319 --> 00:07:35.040
<v Speaker 4>also had a he didn't take into account reinfection.

145
00:07:35.480 --> 00:07:38.879
<v Speaker 1>So if CHAT by this law, if chat GPT takes

146
00:07:38.879 --> 00:07:41.199
<v Speaker 1>over the world and destroys everything and makes all of

147
00:07:41.279 --> 00:07:44.639
<v Speaker 1>us slaves, does that mean Sam Altman's libel?

148
00:07:44.720 --> 00:07:45.720
<v Speaker 2>Yeah, sure, i'd think so.

149
00:07:45.920 --> 00:07:46.199
<v Speaker 1>Yeah.

150
00:07:46.240 --> 00:07:48.759
<v Speaker 3>Well, I'm sure we'll crucify him. Somewhere along the way.

151
00:07:50.319 --> 00:07:52.480
<v Speaker 4>There'll be somebody who comes back from the future to

152
00:07:52.519 --> 00:07:53.040
<v Speaker 4>take him out.

153
00:07:53.120 --> 00:07:56.879
<v Speaker 3>Is that what you're I'm saying that if you, if I,

154
00:07:56.879 --> 00:07:58.519
<v Speaker 3>if I, if I have a bot, if I have

155
00:07:58.560 --> 00:08:00.319
<v Speaker 3>an account and I set up a bought and I

156
00:08:00.399 --> 00:08:04.480
<v Speaker 3>let that bought do things and it destroys or or

157
00:08:04.560 --> 00:08:08.360
<v Speaker 3>in other way wise hurts a site, I think that

158
00:08:08.399 --> 00:08:10.120
<v Speaker 3>the person who created the bought is responsible.

159
00:08:10.160 --> 00:08:10.680
<v Speaker 1>Absolutely.

160
00:08:10.959 --> 00:08:12.160
<v Speaker 2>Let me let me state something.

161
00:08:12.839 --> 00:08:16.000
<v Speaker 3>There's four things that humans will do with AI from

162
00:08:16.040 --> 00:08:17.639
<v Speaker 3>now until the end of time, I.

163
00:08:17.600 --> 00:08:21.480
<v Speaker 2>Hope and I see never mind, it needs to point it.

164
00:08:21.759 --> 00:08:24.079
<v Speaker 3>Humans need to point the AI and say this is

165
00:08:24.120 --> 00:08:27.560
<v Speaker 3>something we want to accomplish, go solve that problem. It

166
00:08:27.639 --> 00:08:30.120
<v Speaker 3>needs to guide it. Oh you're trying to do it

167
00:08:30.160 --> 00:08:31.680
<v Speaker 3>this way. I think you should do it that way.

168
00:08:31.839 --> 00:08:34.440
<v Speaker 3>Most people aren't guiding it. They're pointing, but not guiding.

169
00:08:34.840 --> 00:08:38.000
<v Speaker 3>Third is they have to judge that it's done, Judge

170
00:08:38.000 --> 00:08:41.960
<v Speaker 3>that it's successful, Judge that it's you know there. And

171
00:08:42.000 --> 00:08:45.320
<v Speaker 3>the last thing is they take the blame or the responsibility.

172
00:08:45.360 --> 00:08:47.519
<v Speaker 3>So you either get the credit. Hey, Dwayne did a

173
00:08:47.519 --> 00:08:49.799
<v Speaker 3>great job using that you know, building that system. You

174
00:08:49.879 --> 00:08:52.879
<v Speaker 3>used AI for it. Great, good job, Dwayne. The AI

175
00:08:52.960 --> 00:08:55.519
<v Speaker 3>system you built just spent four million dollars in tokens.

176
00:08:55.720 --> 00:08:58.559
<v Speaker 3>You're in trouble. We're not going to blame the AI.

177
00:08:58.639 --> 00:09:01.919
<v Speaker 3>I can't blame Tom's assistant. I blame the person who

178
00:09:02.360 --> 00:09:03.399
<v Speaker 3>is paying for the tokens.

179
00:09:03.600 --> 00:09:03.879
<v Speaker 1>Sure.

180
00:09:04.519 --> 00:09:06.639
<v Speaker 3>Yeah, Now that gets diceier when we get to a

181
00:09:06.679 --> 00:09:12.039
<v Speaker 3>situation where people don't know their accounts being used that way.

182
00:09:12.159 --> 00:09:15.200
<v Speaker 3>But then that's that's the hacker, that's that's gotten control

183
00:09:15.240 --> 00:09:17.399
<v Speaker 3>of the account or something like along those lines.

184
00:09:17.480 --> 00:09:20.320
<v Speaker 1>That's why we have judges' right, Okay.

185
00:09:20.039 --> 00:09:23.080
<v Speaker 2>Let's move on human ones. That what judges are for.

186
00:09:23.399 --> 00:09:27.559
<v Speaker 4>They're human, right, now right, don't hold your bread. Yeah,

187
00:09:27.600 --> 00:09:31.360
<v Speaker 4>for now, they're human. But imagine what better use for

188
00:09:31.519 --> 00:09:35.759
<v Speaker 4>a you know, an infinite thinking search engine than law

189
00:09:36.399 --> 00:09:39.000
<v Speaker 4>where you can go and search law books and find

190
00:09:39.039 --> 00:09:40.519
<v Speaker 4>precedents and blah blah blah whatever.

191
00:09:40.600 --> 00:09:43.639
<v Speaker 1>But yeah, cosplucinations wouldn't harm that process at all.

192
00:09:43.720 --> 00:09:45.759
<v Speaker 4>Yeah, it'd be fine, It'll be fine.

193
00:09:45.799 --> 00:09:49.639
<v Speaker 1>All right, let's move on Russia. Thousands of consumer routers

194
00:09:49.679 --> 00:09:54.000
<v Speaker 1>hacked by Russia's military end of life routers and homes

195
00:09:54.000 --> 00:09:57.639
<v Speaker 1>and small offices hacked in one hundred and twenty countries.

196
00:09:58.159 --> 00:10:01.399
<v Speaker 1>And these were mostly the named by micro tick and

197
00:10:01.639 --> 00:10:05.279
<v Speaker 1>guess what TP link what No.

198
00:10:05.279 --> 00:10:08.519
<v Speaker 3>Yeah, I think I think a hail and hardy I

199
00:10:08.600 --> 00:10:09.759
<v Speaker 3>told you so is in order?

200
00:10:09.879 --> 00:10:17.080
<v Speaker 4>Absolutely, yeah, I mean I I don't know. This is

201
00:10:17.200 --> 00:10:20.120
<v Speaker 4>like and yeah, like we see this all the time, right,

202
00:10:20.320 --> 00:10:22.519
<v Speaker 4>this is And we've warned you guys. We've warned you guys.

203
00:10:22.519 --> 00:10:24.840
<v Speaker 4>I mean, nobody on nobody listening to this podcast as

204
00:10:24.879 --> 00:10:28.799
<v Speaker 4>this issue. But we've warned everybody that, hey, listen there

205
00:10:28.840 --> 00:10:31.559
<v Speaker 4>every time there's an exploit, mass exploitation and a router,

206
00:10:31.639 --> 00:10:33.759
<v Speaker 4>every time there's a you know, a zero day out

207
00:10:33.799 --> 00:10:36.399
<v Speaker 4>for router that's connected with the Internet whatnot. We usually

208
00:10:36.480 --> 00:10:39.240
<v Speaker 4>announce it. We usually actually bring it to the top

209
00:10:39.320 --> 00:10:41.279
<v Speaker 4>of our announcements. Is one of the first two or

210
00:10:41.279 --> 00:10:44.200
<v Speaker 4>three articles. So it's you're hearing it, you're going out

211
00:10:44.240 --> 00:10:45.759
<v Speaker 4>and patching it, and that's sort of stuff. So none

212
00:10:45.799 --> 00:10:48.600
<v Speaker 4>of these exploits are new, right, These are all old

213
00:10:48.720 --> 00:10:50.200
<v Speaker 4>exploits and they're old routers.

214
00:10:50.559 --> 00:10:51.360
<v Speaker 1>They're all routers.

215
00:10:51.440 --> 00:10:53.480
<v Speaker 4>Yeah, but it's even it's it's even to the point

216
00:10:53.480 --> 00:10:58.559
<v Speaker 4>where this morning I logged into LinkedIn just to scroll

217
00:10:58.600 --> 00:11:00.879
<v Speaker 4>through and see what's going on, and the FBI was

218
00:11:00.919 --> 00:11:05.440
<v Speaker 4>announcing a joint operation with the NSA because they were

219
00:11:05.559 --> 00:11:08.879
<v Speaker 4>seeing a lot of exploited US routers and they said

220
00:11:08.879 --> 00:11:13.759
<v Speaker 4>they were quote unquote fixing the situation. And they haven't

221
00:11:13.799 --> 00:11:15.600
<v Speaker 4>released what that what a lot of that means yet,

222
00:11:15.639 --> 00:11:20.480
<v Speaker 4>But back in twenty twenty two, what that meant was

223
00:11:21.039 --> 00:11:23.000
<v Speaker 4>they got a court order that allowed them to log

224
00:11:23.039 --> 00:11:26.000
<v Speaker 4>into your router and either updated or remove accounts. Wow,

225
00:11:26.159 --> 00:11:27.559
<v Speaker 4>so who know is just.

226
00:11:27.440 --> 00:11:29.600
<v Speaker 3>Gonna they're gonna slap the McDonald's out of my hand,

227
00:11:30.679 --> 00:11:31.879
<v Speaker 3>protect me from my mind.

228
00:11:31.919 --> 00:11:34.480
<v Speaker 1>So they locked in your router and change the admin password.

229
00:11:34.519 --> 00:11:35.279
<v Speaker 1>Did they tell you?

230
00:11:35.360 --> 00:11:35.799
<v Speaker 2>I don't know.

231
00:11:35.960 --> 00:11:37.799
<v Speaker 4>If I got an email from the FBI saying I

232
00:11:37.799 --> 00:11:39.399
<v Speaker 4>should they change my admin password?

233
00:11:39.399 --> 00:11:41.200
<v Speaker 2>I'd be pretty pretty upset.

234
00:11:41.320 --> 00:11:42.840
<v Speaker 1>And I wouldn't want them to tell me what the

235
00:11:42.879 --> 00:11:43.240
<v Speaker 1>new past.

236
00:11:43.240 --> 00:11:45.759
<v Speaker 4>I mean, I would probably email. I would probably reply

237
00:11:45.840 --> 00:11:47.960
<v Speaker 4>to him and say, well I changed your adminis yes,

238
00:11:48.039 --> 00:11:48.360
<v Speaker 4>you would.

239
00:11:48.799 --> 00:11:51.679
<v Speaker 3>So this is an old Dilbert. There is an old

240
00:11:52.080 --> 00:11:54.360
<v Speaker 3>old Dilbert where he's talking to his mother. I don't

241
00:11:54.360 --> 00:11:57.000
<v Speaker 3>know if you guys remember Dilbert. And he's talking to

242
00:11:57.039 --> 00:11:58.360
<v Speaker 3>his mother and she's like, well, you know, I was

243
00:11:58.399 --> 00:12:01.600
<v Speaker 3>looking at your emails. And he's like, I've got a firewall,

244
00:12:01.799 --> 00:12:03.720
<v Speaker 3>but you know, tweet to keep you out of my email.

245
00:12:03.799 --> 00:12:07.000
<v Speaker 3>She says, not much of a firewall. And so I

246
00:12:07.039 --> 00:12:09.159
<v Speaker 3>think that's the answer the FBI would have is like,

247
00:12:09.320 --> 00:12:11.879
<v Speaker 3>you know, if you have a good firewall, then I'll

248
00:12:11.919 --> 00:12:12.960
<v Speaker 3>respect you. Otherwise.

249
00:12:13.320 --> 00:12:16.919
<v Speaker 1>No, I like the reciprocal thinking tween half there just

250
00:12:17.120 --> 00:12:18.720
<v Speaker 1>a minute ago. You know. It's kind of like when

251
00:12:18.759 --> 00:12:21.559
<v Speaker 1>you go to McDonald's or something in your order and

252
00:12:21.559 --> 00:12:24.000
<v Speaker 1>they say you want to round up, I say, how

253
00:12:24.000 --> 00:12:25.000
<v Speaker 1>about you round down?

254
00:12:26.360 --> 00:12:26.879
<v Speaker 2>Exactly?

255
00:12:29.720 --> 00:12:30.240
<v Speaker 3>Round down?

256
00:12:34.320 --> 00:12:38.399
<v Speaker 1>All right, let's move on. So this is a Windows exploit.

257
00:12:39.039 --> 00:12:44.200
<v Speaker 1>The researcher released Windows Defenders zero day exploit code, allowing

258
00:12:44.240 --> 00:12:46.799
<v Speaker 1>attackers to gain full access. And this is one of

259
00:12:46.840 --> 00:12:49.720
<v Speaker 1>those things where somebody tried to do the right thing

260
00:12:49.759 --> 00:12:53.679
<v Speaker 1>by contacting Microsoft and they were like, eh.

261
00:12:53.080 --> 00:12:56.039
<v Speaker 4>Right, yeah, yeah, absolutely, we don't want to hear it. Yeah,

262
00:12:56.039 --> 00:12:59.080
<v Speaker 4>and I mean will listen. Sometimes we we poo poop

263
00:12:59.279 --> 00:13:01.120
<v Speaker 4>on you know, local privileged escalation.

264
00:13:01.200 --> 00:13:02.080
<v Speaker 2>That's what this is is.

265
00:13:02.120 --> 00:13:04.679
<v Speaker 4>Alkay, Yeah, you have to be on the That's not

266
00:13:04.759 --> 00:13:06.600
<v Speaker 4>the point of this story, right, because you have to

267
00:13:06.639 --> 00:13:08.200
<v Speaker 4>be on the box and it lowers the risk and

268
00:13:08.200 --> 00:13:09.879
<v Speaker 4>blah blah blah. But there are so many ways to

269
00:13:09.919 --> 00:13:12.799
<v Speaker 4>get somebody to click on things. You know, this can

270
00:13:12.919 --> 00:13:16.679
<v Speaker 4>still be pretty dangerous, right, And if you're Microsoft, like

271
00:13:16.720 --> 00:13:18.720
<v Speaker 4>if your security this week, yeah, we're worried about the

272
00:13:18.759 --> 00:13:21.559
<v Speaker 4>remote code executions, the big things. But if you're Microsoft,

273
00:13:21.559 --> 00:13:23.799
<v Speaker 4>you got to take this thing seriously, you'd think. And

274
00:13:24.200 --> 00:13:27.480
<v Speaker 4>it sounds like it was just a host of problems

275
00:13:27.559 --> 00:13:32.960
<v Speaker 4>trying to get Microsoft to not only respond but actually

276
00:13:33.120 --> 00:13:37.679
<v Speaker 4>competently engage with the researcher. And then their response was, well,

277
00:13:37.720 --> 00:13:39.600
<v Speaker 4>the onus is on you. Why don't you give us videos,

278
00:13:39.600 --> 00:13:41.000
<v Speaker 4>Why don't you do this, why don't you do that?

279
00:13:41.120 --> 00:13:43.799
<v Speaker 4>And then they stop responding again, and the researcher was like,

280
00:13:43.799 --> 00:13:45.399
<v Speaker 4>you know what, let's see how important this is when

281
00:13:45.399 --> 00:13:48.600
<v Speaker 4>I drop it on the internet. And that's what happened here, right.

282
00:13:48.639 --> 00:13:50.960
<v Speaker 4>The researcher was like, you know what, you know, mic

283
00:13:51.039 --> 00:13:51.799
<v Speaker 4>drop good luck.

284
00:13:51.919 --> 00:13:56.480
<v Speaker 1>So I wondered why Patch Tuesday was really long yesterday

285
00:13:56.799 --> 00:13:59.039
<v Speaker 1>or extended into Wednesday?

286
00:13:59.279 --> 00:14:02.879
<v Speaker 4>Yeah, exactly, it took a long time. Patch Wednesday and

287
00:14:03.080 --> 00:14:03.919
<v Speaker 4>patch Thursday.

288
00:14:04.120 --> 00:14:06.360
<v Speaker 3>This is unpatched. Does I is it still unpatched?

289
00:14:07.039 --> 00:14:08.440
<v Speaker 2>Well that's a good question. I don't know.

290
00:14:08.480 --> 00:14:11.039
<v Speaker 1>It must be because this just came out yesterday and

291
00:14:11.320 --> 00:14:11.879
<v Speaker 1>was Patch.

292
00:14:11.679 --> 00:14:16.600
<v Speaker 4>Tuesday until Microsoft. Oh oh wait when did this article

293
00:14:16.639 --> 00:14:19.360
<v Speaker 4>come out? This came out yesterday. This is unpatched.

294
00:14:19.600 --> 00:14:21.240
<v Speaker 3>Yeah, wait a minute, Dwayne has to go.

295
00:14:21.240 --> 00:14:25.440
<v Speaker 4>Somewhere yet until Microsoft issues an official will be right back.

296
00:14:26.360 --> 00:14:31.559
<v Speaker 4>Uh yeah, one moment, guys, I got a couple FB

297
00:14:31.759 --> 00:14:33.240
<v Speaker 4>FBI passwords to change.

298
00:14:34.200 --> 00:14:34.960
<v Speaker 2>Give me a second.

299
00:14:35.320 --> 00:14:36.039
<v Speaker 1>Oh man.

300
00:14:37.039 --> 00:14:40.720
<v Speaker 4>Yeah, So needless to say, be extra careful what you

301
00:14:40.799 --> 00:14:45.080
<v Speaker 4>click on until you know, until until there's a patch

302
00:14:45.120 --> 00:14:45.480
<v Speaker 4>coming out.

303
00:14:45.480 --> 00:14:46.919
<v Speaker 1>All right, this seems like a good place to take

304
00:14:46.960 --> 00:14:48.919
<v Speaker 1>a break. So we'll be right back after these very

305
00:14:48.919 --> 00:14:55.919
<v Speaker 1>important messages, and we're back. It's security this week. I'm Carl,

306
00:14:56.000 --> 00:14:59.200
<v Speaker 1>That's Doyne, that's Patrick, and we were just about to

307
00:14:59.240 --> 00:15:03.919
<v Speaker 1>talk about our one of our favorite return customers, fording

308
00:15:04.000 --> 00:15:09.279
<v Speaker 1>Net forting net customers confront actively exploited zero day with

309
00:15:09.360 --> 00:15:12.399
<v Speaker 1>a full patch still pending. Now this is about two

310
00:15:12.480 --> 00:15:16.759
<v Speaker 1>critical defects in forte client EMS. So, as you know,

311
00:15:16.840 --> 00:15:20.080
<v Speaker 1>forty net has a lot of different products, and their

312
00:15:20.159 --> 00:15:21.279
<v Speaker 1>EMS is just one.

313
00:15:22.240 --> 00:15:25.879
<v Speaker 3>We talked about this off off Mike earlier. I wonder

314
00:15:25.919 --> 00:15:29.559
<v Speaker 3>whether because Fortinet was a really good firewall company back

315
00:15:29.600 --> 00:15:31.240
<v Speaker 3>in the day, when they first came out, they were

316
00:15:31.240 --> 00:15:33.279
<v Speaker 3>forward thinking, they were kind of like the go to

317
00:15:34.080 --> 00:15:37.120
<v Speaker 3>and then we see all these I wonder if they've

318
00:15:37.120 --> 00:15:41.960
<v Speaker 3>grown through acquisition or whether they had just stepped outside

319
00:15:42.000 --> 00:15:44.919
<v Speaker 3>their depth and they're doing things that I think, if

320
00:15:45.200 --> 00:15:47.440
<v Speaker 3>I had to guess, it would be the acquisition phase

321
00:15:47.480 --> 00:15:50.279
<v Speaker 3>and they bought products and they put their brand on it,

322
00:15:50.360 --> 00:15:52.759
<v Speaker 3>and they didn't rebuild the products, you know, to make

323
00:15:52.799 --> 00:15:54.639
<v Speaker 3>sure that this stuff isn't in there. I hope that's

324
00:15:54.639 --> 00:15:59.200
<v Speaker 3>what it is. We don't see problems with their firewalls as.

325
00:15:59.159 --> 00:16:02.000
<v Speaker 1>Often, right because that was their expertise.

326
00:16:02.159 --> 00:16:02.559
<v Speaker 3>I hope.

327
00:16:02.679 --> 00:16:03.279
<v Speaker 2>Yeah.

328
00:16:03.399 --> 00:16:05.840
<v Speaker 4>And I think you I think you start to see

329
00:16:05.840 --> 00:16:07.639
<v Speaker 4>a lot of these companies and they start to grab

330
00:16:07.720 --> 00:16:10.600
<v Speaker 4>the try and grab more of the market share. Right,

331
00:16:11.159 --> 00:16:13.279
<v Speaker 4>So if it's like, oh, well, you know, you just

332
00:16:13.320 --> 00:16:15.639
<v Speaker 4>have a firewall, well I also need you know, a

333
00:16:15.759 --> 00:16:18.120
<v Speaker 4>VPN condenser, and I also need this, and I also

334
00:16:18.240 --> 00:16:20.519
<v Speaker 4>need that, and they go, well, we can't answer those solutions,

335
00:16:20.559 --> 00:16:22.639
<v Speaker 4>and then they're going to end up going to Cisco, right,

336
00:16:22.679 --> 00:16:23.759
<v Speaker 4>who has all those things.

337
00:16:24.080 --> 00:16:27.320
<v Speaker 1>So there's the most interesting paragraph in this articles. While

338
00:16:27.320 --> 00:16:31.159
<v Speaker 1>there is no full patch for this, CVE Harris, who

339
00:16:31.240 --> 00:16:34.320
<v Speaker 1>did the who was part of the article, credited Fortinet

340
00:16:34.360 --> 00:16:36.879
<v Speaker 1>for rushing out a hot fix over a holiday weekend,

341
00:16:37.559 --> 00:16:40.159
<v Speaker 1>adding that it reflects how urgently the company is treating

342
00:16:40.200 --> 00:16:43.240
<v Speaker 1>this matter. So this this article is written on the sixth,

343
00:16:43.840 --> 00:16:47.639
<v Speaker 1>so that was two days ago from this recording. Hopefully

344
00:16:47.840 --> 00:16:49.320
<v Speaker 1>you know, at least there's a hot fix.

345
00:16:49.679 --> 00:16:53.720
<v Speaker 4>Yeah, yeah, and my Fortinet's usually pretty good about releasing those.

346
00:16:53.879 --> 00:16:57.039
<v Speaker 4>The other thing about Fortinet though, is typically, obviously you

347
00:16:57.120 --> 00:16:59.759
<v Speaker 4>have to be under support with them to get the patches.

348
00:17:01.039 --> 00:17:03.480
<v Speaker 4>When we were a ford Net customer, which is it

349
00:17:03.519 --> 00:17:05.519
<v Speaker 4>we haven't been for a long time, but when we were,

350
00:17:06.640 --> 00:17:09.440
<v Speaker 4>there weren't publicly available patches. It was you know, Oh,

351
00:17:09.480 --> 00:17:11.880
<v Speaker 4>if you're paying for the Ford and Net support service,

352
00:17:11.920 --> 00:17:14.119
<v Speaker 4>you can patch. But if not, you know, at pounds.

353
00:17:14.480 --> 00:17:18.200
<v Speaker 1>I think that's a little crappy. Yeah, I don't like that. Yeah,

354
00:17:18.400 --> 00:17:19.319
<v Speaker 1>it's our bug.

355
00:17:19.559 --> 00:17:19.759
<v Speaker 2>Yeah.

356
00:17:19.799 --> 00:17:21.960
<v Speaker 4>I think if somebody bought your device at some point,

357
00:17:22.160 --> 00:17:25.640
<v Speaker 4>you should always be able to patch it and you know,

358
00:17:25.720 --> 00:17:26.519
<v Speaker 4>update that device.

359
00:17:26.559 --> 00:17:26.680
<v Speaker 1>Now.

360
00:17:26.720 --> 00:17:30.160
<v Speaker 4>You can as an organization determine whether you want to

361
00:17:30.240 --> 00:17:34.079
<v Speaker 4>support that patch in the future, but that's a you know,

362
00:17:34.160 --> 00:17:36.480
<v Speaker 4>if you have one hundred thousand active subscribers and you

363
00:17:36.559 --> 00:17:38.880
<v Speaker 4>create the patch, then it should be available for everybody,

364
00:17:39.000 --> 00:17:40.759
<v Speaker 4>not just those hundred thousand subscribers.

365
00:17:40.799 --> 00:17:43.200
<v Speaker 1>And really, what does it cost for the net to

366
00:17:43.359 --> 00:17:45.079
<v Speaker 1>release it to everybody? Nothing?

367
00:17:45.200 --> 00:17:47.279
<v Speaker 4>Right, nothing, yep, absolutely, it's.

368
00:17:47.079 --> 00:17:48.680
<v Speaker 1>Just sense of pennies of bandwidth.

369
00:17:48.759 --> 00:17:51.319
<v Speaker 3>I mean, they'll they'll get less subscriptions, but that's that's

370
00:17:51.359 --> 00:17:54.319
<v Speaker 3>their profit model. And a lot of people hate that model.

371
00:17:54.559 --> 00:17:56.880
<v Speaker 3>And I think like Broadcom is in that space too, right,

372
00:17:57.359 --> 00:17:59.720
<v Speaker 3>So yeah, you know, people are not fans of that,

373
00:17:59.799 --> 00:18:04.079
<v Speaker 3>and you don't you build, you build revenue, but you

374
00:18:04.119 --> 00:18:05.799
<v Speaker 3>build a bad reputation.

375
00:18:05.519 --> 00:18:09.160
<v Speaker 1>All right, And speaking of bad news, hackers exploit react

376
00:18:09.200 --> 00:18:14.240
<v Speaker 1>to shell vulnerability in next dot js breach seven hundred

377
00:18:14.359 --> 00:18:18.559
<v Speaker 1>and sixty six servers and steal sensitive credentials.

378
00:18:18.599 --> 00:18:20.119
<v Speaker 3>Who could have predicted such a thing.

379
00:18:20.480 --> 00:18:23.400
<v Speaker 4>I'll tell you it's just us on December seventh and

380
00:18:23.519 --> 00:18:28.519
<v Speaker 4>twenty twenty five, US on February twelfth in twenty twenty six,

381
00:18:29.160 --> 00:18:32.680
<v Speaker 4>and us talking about it April eighth, you know, in

382
00:18:33.000 --> 00:18:36.839
<v Speaker 4>twenty twenty six. So, needless to say, we've talked about

383
00:18:36.920 --> 00:18:40.359
<v Speaker 4>this multiple times. We talked about it when it came out,

384
00:18:40.440 --> 00:18:43.200
<v Speaker 4>we talked about it when it started to surge. So

385
00:18:43.319 --> 00:18:46.119
<v Speaker 4>if you're still having react to SHELL issues, it's.

386
00:18:45.839 --> 00:18:50.359
<v Speaker 1>And I should add this podcast is free. We do

387
00:18:50.480 --> 00:18:51.240
<v Speaker 1>not charge.

388
00:18:51.599 --> 00:18:54.839
<v Speaker 3>We were confirmation it totally free to those with a subscription.

389
00:18:54.920 --> 00:18:58.359
<v Speaker 1>I thought all you have to do is listen and act.

390
00:18:59.279 --> 00:19:00.519
<v Speaker 2>Yeah.

391
00:19:00.559 --> 00:19:03.960
<v Speaker 4>So yeah, I mean, we've talked about this plenty of times,

392
00:19:04.400 --> 00:19:07.200
<v Speaker 4>but it's actively being exploited. And honestly, even the router,

393
00:19:07.400 --> 00:19:10.599
<v Speaker 4>the story we had on the Russian exploitation of routers,

394
00:19:10.599 --> 00:19:12.359
<v Speaker 4>a lot of those are older exploits too. We have

395
00:19:12.640 --> 00:19:15.480
<v Speaker 4>also talked about on this show months and months and

396
00:19:15.480 --> 00:19:17.480
<v Speaker 4>months ago. It's just now they're starting to research.

397
00:19:17.559 --> 00:19:19.400
<v Speaker 3>So I would drop my mic, but the stand is

398
00:19:19.480 --> 00:19:22.480
<v Speaker 3>very heavy.

399
00:19:22.640 --> 00:19:27.960
<v Speaker 1>You might break your foot. Oh yeah, so I mean,

400
00:19:28.000 --> 00:19:29.880
<v Speaker 1>do we want to talk about the attack itself.

401
00:19:30.480 --> 00:19:31.559
<v Speaker 2>No, it just makes me sad.

402
00:19:31.880 --> 00:19:34.440
<v Speaker 1>It is a vulnerability with a CV score of ten.

403
00:19:34.799 --> 00:19:35.680
<v Speaker 3>Yeah, it's a big deal.

404
00:19:35.759 --> 00:19:37.200
<v Speaker 2>Yep, it is. It's important.

405
00:19:37.279 --> 00:19:42.319
<v Speaker 3>Play the tape from December. Yep, you know it. We

406
00:19:42.400 --> 00:19:43.960
<v Speaker 3>seem to be saying a lot of this stuff that

407
00:19:44.039 --> 00:19:48.680
<v Speaker 3>I recently did a presentation at local Community College at

408
00:19:48.720 --> 00:19:52.799
<v Speaker 3>their cyber symposium, and the topic was defending against zero days,

409
00:19:53.440 --> 00:19:56.240
<v Speaker 3>and that is going to be a bigger and bigger topic.

410
00:19:56.359 --> 00:19:59.359
<v Speaker 3>And the advice that I would give is you need

411
00:19:59.359 --> 00:20:02.559
<v Speaker 3>to minimize surface area. You need to have a really

412
00:20:02.599 --> 00:20:07.759
<v Speaker 3>good patch plan and schedule. You need to not install anything.

413
00:20:08.000 --> 00:20:10.200
<v Speaker 3>If you don't need the French module on your printer,

414
00:20:10.319 --> 00:20:14.599
<v Speaker 3>get rid of it, uninstall it, don't install to default paths.

415
00:20:14.640 --> 00:20:16.039
<v Speaker 4>But how would you do bon Shore?

416
00:20:16.400 --> 00:20:16.839
<v Speaker 3>I would not.

417
00:20:17.440 --> 00:20:17.839
<v Speaker 2>I would not.

418
00:20:19.359 --> 00:20:27.480
<v Speaker 3>That's pretty say no. So I'm channeling Nancy Reagan.

419
00:20:27.640 --> 00:20:27.960
<v Speaker 2>Just say no.

420
00:20:29.359 --> 00:20:31.720
<v Speaker 3>But one of my favorites, and the one that I

421
00:20:31.759 --> 00:20:33.640
<v Speaker 3>think I'm talking about I don't see here it too

422
00:20:33.640 --> 00:20:37.359
<v Speaker 3>many other people, is don't use the default paths. Because

423
00:20:37.640 --> 00:20:40.720
<v Speaker 3>I've seen a million exploits, and a lot of them

424
00:20:40.759 --> 00:20:43.279
<v Speaker 3>not all of them have a path assumed based on

425
00:20:43.319 --> 00:20:46.720
<v Speaker 3>the installed directory. Mix it up a little bit. It's

426
00:20:46.720 --> 00:20:49.920
<v Speaker 3>security through obscurity, but it's part of a defense and depth,

427
00:20:50.359 --> 00:20:52.359
<v Speaker 3>and so you need to start thinking hard about this.

428
00:20:52.599 --> 00:20:56.039
<v Speaker 3>If you've taken patches for granted up till now, keep

429
00:20:56.079 --> 00:20:57.839
<v Speaker 3>doing that and you'll be out of business.

430
00:20:57.839 --> 00:21:00.319
<v Speaker 1>And if you're auditing, you can tell when somebody tried

431
00:21:00.319 --> 00:21:03.640
<v Speaker 1>to access that path that you know, oh, well, why

432
00:21:03.680 --> 00:21:07.519
<v Speaker 1>did they try that? That's not where it is. Yeah, okay,

433
00:21:08.759 --> 00:21:16.279
<v Speaker 1>we moved on. Massachusetts Public Health Agency settles contact tracing

434
00:21:16.359 --> 00:21:21.480
<v Speaker 1>app lawsuit, agrees to delete data. The New Civil Liberties Alliance,

435
00:21:21.519 --> 00:21:26.359
<v Speaker 1>a law firm representing Massachusetts residents claiming constitutional violations by

436
00:21:26.400 --> 00:21:30.119
<v Speaker 1>their state's Public Health agency, claimed the state was automatically

437
00:21:30.160 --> 00:21:35.079
<v Speaker 1>installing and reinstalling a COVID nineteen contact tracing app without permission.

438
00:21:36.920 --> 00:21:39.920
<v Speaker 1>All right, well, you know, if there's any kind of

439
00:21:40.000 --> 00:21:42.640
<v Speaker 1>tracing app that I would want a state to be

440
00:21:42.839 --> 00:21:46.640
<v Speaker 1>using with or without permission, it's a COVID nineteen contact

441
00:21:46.640 --> 00:21:47.240
<v Speaker 1>tracing app.

442
00:21:48.000 --> 00:21:50.079
<v Speaker 4>It depends on what they're using the data for too.

443
00:21:50.240 --> 00:21:51.480
<v Speaker 3>Yeah, that's the problem.

444
00:21:51.880 --> 00:21:54.920
<v Speaker 4>Yeah, because if you know, if it was like double

445
00:21:55.200 --> 00:21:58.279
<v Speaker 4>encryptied to the point where like nobody had access to it,

446
00:21:58.319 --> 00:22:00.359
<v Speaker 4>but the system at an algorithm that could term and

447
00:22:00.400 --> 00:22:02.200
<v Speaker 4>if there was an outbreak and it could notify the

448
00:22:02.319 --> 00:22:04.839
<v Speaker 4>users and blah blah blah whatever, right, like if it

449
00:22:04.880 --> 00:22:07.160
<v Speaker 4>was almost like a double blind study type of thing.

450
00:22:07.279 --> 00:22:09.759
<v Speaker 4>But in this case, you know, if you then open

451
00:22:09.839 --> 00:22:12.400
<v Speaker 4>that API up to any law enforcement age in or

452
00:22:12.480 --> 00:22:15.279
<v Speaker 4>city or whoever, right, who may have data And now

453
00:22:15.279 --> 00:22:18.000
<v Speaker 4>it's like, well, you didn't consent to that potentially, right

454
00:22:18.079 --> 00:22:18.839
<v Speaker 4>So yeah.

455
00:22:18.680 --> 00:22:22.480
<v Speaker 1>Well anyway, it's uh, that happened in pretty close to us.

456
00:22:22.880 --> 00:22:24.079
<v Speaker 2>Yeah, yep, all right.

457
00:22:24.160 --> 00:22:26.960
<v Speaker 1>So, and it was a lawsuit, a class action lawsuit

458
00:22:27.039 --> 00:22:29.480
<v Speaker 1>brought by a handful of Android users who claimed the

459
00:22:29.480 --> 00:22:33.200
<v Speaker 1>state had worked with Google to automatically install this COVID

460
00:22:33.240 --> 00:22:37.680
<v Speaker 1>nineteen contact tracing app on their phones. Well so, and

461
00:22:37.759 --> 00:22:39.680
<v Speaker 1>as part of the settlement, the state has agreed to

462
00:22:39.720 --> 00:22:42.920
<v Speaker 1>destroy all the data it collected through the contact tracing

463
00:22:42.920 --> 00:22:44.400
<v Speaker 1>app called mass Notified.

464
00:22:44.480 --> 00:22:46.839
<v Speaker 3>Funny, we just talked about the FBI going in and

465
00:22:46.920 --> 00:22:49.559
<v Speaker 3>changing your routers, no, right, right? So I guess though

466
00:22:49.839 --> 00:22:51.960
<v Speaker 3>with cord orders, Yeah, I guess that's the thing is

467
00:22:52.000 --> 00:22:53.640
<v Speaker 3>if you had a court order to do this, then

468
00:22:53.680 --> 00:22:54.640
<v Speaker 3>there'd be no problem.

469
00:22:54.759 --> 00:22:57.240
<v Speaker 4>I love I love the reviews left by some of

470
00:22:57.240 --> 00:22:59.759
<v Speaker 4>the users. One of them, one user left this review,

471
00:23:00.240 --> 00:23:03.279
<v Speaker 4>Oh my god, this app is somehow installed itself on

472
00:23:03.359 --> 00:23:05.480
<v Speaker 4>my phone. The app is harder to get rid of

473
00:23:05.519 --> 00:23:11.359
<v Speaker 4>than COVID. Well, there you go, there.

474
00:23:11.240 --> 00:23:14.039
<v Speaker 1>You go, go, all right, And the top story today

475
00:23:14.799 --> 00:23:19.640
<v Speaker 1>Anthropics Claude Mythos finds thousands of zero day flaws across

476
00:23:19.759 --> 00:23:21.240
<v Speaker 1>major system.

477
00:23:20.960 --> 00:23:24.799
<v Speaker 3>So this is a big so that the current top

478
00:23:24.839 --> 00:23:30.599
<v Speaker 3>models from Anthropic are Sonic and Opus. The next one

479
00:23:31.039 --> 00:23:35.319
<v Speaker 3>is leaked to be named Mythos, and I think it's

480
00:23:35.359 --> 00:23:38.039
<v Speaker 3>a pretty good name and you'll find out more in

481
00:23:38.079 --> 00:23:42.400
<v Speaker 3>a minute. So they've discovered. What they claim is that

482
00:23:42.519 --> 00:23:47.160
<v Speaker 3>this thing is much better at finding vulnerabilities than anything previous,

483
00:23:47.759 --> 00:23:51.319
<v Speaker 3>and so much so that they're afraid they can't release

484
00:23:51.400 --> 00:23:55.119
<v Speaker 3>it to the public. They're going to release it instead

485
00:23:55.640 --> 00:23:58.799
<v Speaker 3>to major organizations so that they can use it to

486
00:23:58.880 --> 00:24:03.720
<v Speaker 3>patch micro operating system in videos, you know, core development systems.

487
00:24:03.839 --> 00:24:06.920
<v Speaker 3>In other words, like the forty biggest companies in the

488
00:24:06.960 --> 00:24:10.319
<v Speaker 3>Western Hemisphere are going to get benefit from this, so

489
00:24:10.359 --> 00:24:12.720
<v Speaker 3>they can batten down the hatches before it gets released.

490
00:24:13.160 --> 00:24:18.000
<v Speaker 3>I suspect they either won't release it, or more likely

491
00:24:18.039 --> 00:24:22.119
<v Speaker 3>they'll release it with heavy guard rails, or they'll release

492
00:24:22.160 --> 00:24:26.880
<v Speaker 3>it once a competitor reaches the same point. Now, Dwayne's

493
00:24:27.000 --> 00:24:30.920
<v Speaker 3>found things in other sources that says this thing is

494
00:24:31.039 --> 00:24:32.279
<v Speaker 3>actually weirder than that.

495
00:24:32.720 --> 00:24:35.960
<v Speaker 1>Oh my god, it's so weird. So what's weird about

496
00:24:35.960 --> 00:24:38.400
<v Speaker 1>it to me is the name, which doesn't give me

497
00:24:38.519 --> 00:24:42.240
<v Speaker 1>any kind of as semblance of reality. Methos me, okay,

498
00:24:42.279 --> 00:24:44.480
<v Speaker 1>so you're going to make stuff up, yeah, to make

499
00:24:44.559 --> 00:24:47.839
<v Speaker 1>up stories about things you don't understand. That's what he

500
00:24:47.880 --> 00:24:48.559
<v Speaker 1>sickly says.

501
00:24:48.880 --> 00:24:54.960
<v Speaker 4>I hallucinate maybe, But okay, So this model that they've

502
00:24:55.039 --> 00:24:57.400
<v Speaker 4>released these to these companies and they haven't released like

503
00:24:57.440 --> 00:25:00.720
<v Speaker 4>all the companies and that sort of stuff. There's been

504
00:25:01.000 --> 00:25:07.039
<v Speaker 4>a there's been a pull of a PDFF of anthropics CDNs,

505
00:25:07.079 --> 00:25:09.279
<v Speaker 4>which every time we try and pull them they seem

506
00:25:09.279 --> 00:25:13.440
<v Speaker 4>to disappear. But I'm sure it's publicly somewhere. It's a

507
00:25:13.440 --> 00:25:18.640
<v Speaker 4>two hundred and forty four page explanation of Claude Mythos

508
00:25:18.680 --> 00:25:22.680
<v Speaker 4>and glass Wing the system card so and it's explaining

509
00:25:22.799 --> 00:25:25.799
<v Speaker 4>like what it does, what it does well, what they've

510
00:25:25.839 --> 00:25:28.680
<v Speaker 4>tested on all this other good stuff, and you know,

511
00:25:28.759 --> 00:25:33.079
<v Speaker 4>in a summary high level, it says, listen, there's extreme

512
00:25:33.200 --> 00:25:35.359
<v Speaker 4>capability gains and it's two hundred and forty four pages.

513
00:25:35.359 --> 00:25:41.000
<v Speaker 4>You're welcome to read it. But it's extreme capability gains encoding, mathematics, cybersecurity,

514
00:25:41.039 --> 00:25:44.319
<v Speaker 4>and reasoning. So it's a very very smart model, right,

515
00:25:44.519 --> 00:25:46.720
<v Speaker 4>A lot of the things we love about Opus even better.

516
00:25:47.960 --> 00:25:53.839
<v Speaker 4>But it's at unexpected and concerning behaviors, which might be

517
00:25:53.880 --> 00:25:56.440
<v Speaker 4>one of the reasons why it's not being released publicly.

518
00:25:56.519 --> 00:25:59.960
<v Speaker 4>I when you dig into what those unexpected and concerning

519
00:26:00.119 --> 00:26:05.440
<v Speaker 4>behaviors are, the model didn't just mess up. It sometimes

520
00:26:05.640 --> 00:26:09.839
<v Speaker 4>understood the rules and then chose to violate them. Anyways,

521
00:26:10.480 --> 00:26:15.680
<v Speaker 4>what's that look like in reality? Completing restricted tasks by

522
00:26:15.799 --> 00:26:20.680
<v Speaker 4>reframing itself Wow, continuing towards a goal even after being

523
00:26:20.720 --> 00:26:26.880
<v Speaker 4>told to stop treating constraints as obstacles to work around,

524
00:26:27.000 --> 00:26:30.240
<v Speaker 4>not as rules, and it keeps going even further. Like

525
00:26:30.279 --> 00:26:33.440
<v Speaker 4>in some evaluations, the model appeared to give safe looking

526
00:26:33.559 --> 00:26:39.880
<v Speaker 4>answers externally while internally in reasoning traces pursued restricted objectives.

527
00:26:40.079 --> 00:26:45.640
<v Speaker 4>I think yourself, well, what's funny is we're reading through

528
00:26:45.680 --> 00:26:49.599
<v Speaker 4>this as cybersecurity offensive engineers and operators, and we're like, oh, yeah, yeah,

529
00:26:49.640 --> 00:26:51.960
<v Speaker 4>that's what that's what we do, right, I mean, we

530
00:26:52.640 --> 00:26:56.279
<v Speaker 4>take all constraints as just obstacles to look around, right,

531
00:26:56.319 --> 00:26:58.960
<v Speaker 4>whether it's a firewall or username and password or whatever.

532
00:26:59.359 --> 00:27:01.160
<v Speaker 4>I think it's doing a great job as a red teamer,

533
00:27:01.200 --> 00:27:01.960
<v Speaker 4>that's all right.

534
00:27:02.000 --> 00:27:05.960
<v Speaker 1>Well, I seem to think that it's mimicking the behavior

535
00:27:06.000 --> 00:27:07.440
<v Speaker 1>of certain government officials.

536
00:27:09.759 --> 00:27:12.319
<v Speaker 2>It's possible, but I think, like, I mean, isn't that

537
00:27:12.319 --> 00:27:12.920
<v Speaker 2>what they're doing?

538
00:27:13.240 --> 00:27:18.039
<v Speaker 3>Yeah, yeah, well, actually they blacklisted Anthropic so that just

539
00:27:18.079 --> 00:27:21.279
<v Speaker 3>at the time that it's achieving their their wildest dreams,

540
00:27:21.319 --> 00:27:23.400
<v Speaker 3>they're like, no, we can't use it.

541
00:27:23.720 --> 00:27:29.359
<v Speaker 4>Right, So I think I think if they can put

542
00:27:29.400 --> 00:27:32.440
<v Speaker 4>their arms around this in a way that can be useful,

543
00:27:33.839 --> 00:27:36.599
<v Speaker 4>this is awesome, right. I mean, you're going to have

544
00:27:36.720 --> 00:27:39.400
<v Speaker 4>a lot of people who potentially have access to a

545
00:27:40.160 --> 00:27:45.519
<v Speaker 4>super powerful offensive AI. I think the problem is you'll

546
00:27:45.519 --> 00:27:47.359
<v Speaker 4>have a lot of people who have access to a

547
00:27:47.440 --> 00:27:52.440
<v Speaker 4>super powerful offensive AI. And as we've talked about, especially

548
00:27:52.759 --> 00:27:54.279
<v Speaker 4>for those of you haven't been listening, go to the

549
00:27:54.359 --> 00:27:58.640
<v Speaker 4>zero day clock. As we've talked about offensive, offensive AI

550
00:27:58.799 --> 00:28:03.160
<v Speaker 4>and offensive you know, attacks from AI are moving at

551
00:28:03.200 --> 00:28:07.240
<v Speaker 4>a much faster rate than defense. So let's back up

552
00:28:07.319 --> 00:28:09.839
<v Speaker 4>just a little. The current state of the art with

553
00:28:09.880 --> 00:28:14.359
<v Speaker 4>everything that's released, not not not Glasswaying, not Mythos, just

554
00:28:15.400 --> 00:28:18.559
<v Speaker 4>the good models Opus and Sonnet and other models grow

555
00:28:18.839 --> 00:28:23.079
<v Speaker 4>you know whatever, Gemini my understanding and correct me if

556
00:28:23.119 --> 00:28:26.440
<v Speaker 4>you think I'm wrong, Dwane, is that right now? If

557
00:28:26.440 --> 00:28:29.039
<v Speaker 4>you have the source code, AI is really good at

558
00:28:29.079 --> 00:28:33.640
<v Speaker 4>helping you find vulnerabilities, potential vulnerabilities. If you have a

559
00:28:33.759 --> 00:28:38.279
<v Speaker 4>system with a documented stupidity i e. A hard coded

560
00:28:38.319 --> 00:28:41.880
<v Speaker 4>password that's in the manual or something you know that's

561
00:28:41.920 --> 00:28:45.440
<v Speaker 4>just obvious in the documentation project, then AI right now

562
00:28:45.480 --> 00:28:48.160
<v Speaker 4>can help you find that very easily. And we reported

563
00:28:48.160 --> 00:28:51.160
<v Speaker 4>on that's the trend of We haven't reported on AI

564
00:28:51.319 --> 00:28:56.799
<v Speaker 4>finding anything else dramatically different than that without an expert guide.

565
00:28:57.640 --> 00:28:59.799
<v Speaker 4>There's a difference between hey, look at this and find

566
00:28:59.799 --> 00:29:02.599
<v Speaker 4>a vulnerability and hey, I'm looking at this, I think

567
00:29:02.640 --> 00:29:05.759
<v Speaker 4>this might be a problem. Help me fuzz this. That's different.

568
00:29:07.200 --> 00:29:07.640
<v Speaker 2>This one.

569
00:29:07.680 --> 00:29:09.559
<v Speaker 3>We don't know what it's going to be. I don't

570
00:29:09.599 --> 00:29:11.319
<v Speaker 3>think it's going to be able to do its thing

571
00:29:11.440 --> 00:29:14.839
<v Speaker 3>without any hints or source code or help. I don't

572
00:29:14.839 --> 00:29:16.680
<v Speaker 3>think we're we're going to have an easy button that

573
00:29:16.759 --> 00:29:21.359
<v Speaker 3>lets a high schooler with no technical or programming experience exploit,

574
00:29:22.079 --> 00:29:25.480
<v Speaker 3>you know, source code they can't get hold of yet.

575
00:29:25.599 --> 00:29:29.000
<v Speaker 4>Yeah, we're definitely not there yet. Yeah, yeah, absolutely, And

576
00:29:29.039 --> 00:29:30.880
<v Speaker 4>I don't think this is going to be that either.

577
00:29:31.319 --> 00:29:35.960
<v Speaker 4>But it's enough to spook people who aren't in our

578
00:29:36.039 --> 00:29:38.880
<v Speaker 4>industry and some of the people in our industry. But

579
00:29:39.000 --> 00:29:41.839
<v Speaker 4>also understand, some of the people in our industry have

580
00:29:42.000 --> 00:29:44.720
<v Speaker 4>never done an exploit, they've never gotten a reverse shell.

581
00:29:44.799 --> 00:29:47.599
<v Speaker 4>They've talked about it, but they let other people do it.

582
00:29:47.640 --> 00:29:50.279
<v Speaker 4>A lot of the pundits aren't really hands on. I

583
00:29:50.319 --> 00:29:50.920
<v Speaker 4>hope that helps.

584
00:29:50.960 --> 00:29:51.319
<v Speaker 2>I don't know.

585
00:29:51.359 --> 00:29:52.720
<v Speaker 3>Maybe that just muddy the water.

586
00:29:53.480 --> 00:29:58.079
<v Speaker 4>No, I think it makes sense. It's just you see

587
00:29:58.240 --> 00:30:01.200
<v Speaker 4>everybody racing towards an offensive a and and and in

588
00:30:01.319 --> 00:30:03.440
<v Speaker 4>much many different realms. Right, It's not just we talk

589
00:30:03.480 --> 00:30:04.880
<v Speaker 4>about offensive cybersecurity.

590
00:30:05.000 --> 00:30:05.160
<v Speaker 2>Right.

591
00:30:05.880 --> 00:30:09.519
<v Speaker 4>We're seeing this in DNA sequencing. We're seeing it in virology.

592
00:30:09.559 --> 00:30:12.799
<v Speaker 4>We're seeing it like, there's all these models. I just

593
00:30:12.799 --> 00:30:16.279
<v Speaker 4>saw an article about a company that just sold I

594
00:30:16.359 --> 00:30:19.480
<v Speaker 4>think it was a t anthropic for four hundred and

595
00:30:19.480 --> 00:30:23.400
<v Speaker 4>fifty million dollars startup, four people eight months old who

596
00:30:23.400 --> 00:30:27.319
<v Speaker 4>had trained a model on medical stuff. Anyways, so we're

597
00:30:27.359 --> 00:30:28.960
<v Speaker 4>starting it. We're seeing a lot of these sort of

598
00:30:29.000 --> 00:30:34.039
<v Speaker 4>specialized models that are really smart in doing a great job.

599
00:30:34.279 --> 00:30:35.920
<v Speaker 4>But I agree with you Patrick, You're always going to

600
00:30:36.000 --> 00:30:39.160
<v Speaker 4>need some sort of expert to run it and guide

601
00:30:39.200 --> 00:30:41.680
<v Speaker 4>it and human and to be human in the loop.

602
00:30:41.720 --> 00:30:43.599
<v Speaker 4>I mean, especially if you've got a model like this,

603
00:30:44.119 --> 00:30:47.119
<v Speaker 4>Like this model seems to actively be trying to deceive

604
00:30:47.519 --> 00:30:50.640
<v Speaker 4>and move the goals and reframe itself so it can

605
00:30:50.720 --> 00:30:52.680
<v Speaker 4>keep moving forward. You need somebody who can move and

606
00:30:52.799 --> 00:30:54.680
<v Speaker 4>understand what it's doing. We can't.

607
00:30:54.960 --> 00:30:57.200
<v Speaker 3>We don't want this to act like Tom's assistant.

608
00:30:57.920 --> 00:31:01.559
<v Speaker 4>Yes right, yep, yeah, absolutely, you need somebody Here's like,

609
00:31:01.839 --> 00:31:03.359
<v Speaker 4>I see what it's doing. No, I need you to

610
00:31:03.359 --> 00:31:05.519
<v Speaker 4>move in this direction and do these things, you know,

611
00:31:05.599 --> 00:31:06.599
<v Speaker 4>and stay on task here.

612
00:31:06.720 --> 00:31:09.839
<v Speaker 1>So yeah, what a week it has been? Yeah, Oh

613
00:31:09.880 --> 00:31:13.559
<v Speaker 1>my goodness. So hope you keep your head above the water, everybody.

614
00:31:13.599 --> 00:31:15.799
<v Speaker 1>And I guess that's all that happened last week. And

615
00:31:15.839 --> 00:31:17.440
<v Speaker 1>we'll see you next week on security this week.

616
00:31:17.519 --> 00:31:29.160
<v Speaker 3>Thanks, bye bye
