1
00:00:06,680 --> 00:00:10,360
Welcome to the Blue Security 
Podcast, a weekly podcast for 

2
00:00:10,360 --> 00:00:13,600
information security defenders 
where we bring you discussions 

3
00:00:13,600 --> 00:00:16,960
on best practices, tools, and 
implementation for enterprise 

4
00:00:16,960 --> 00:00:19,200
security. 
Now here are your hosts for 

5
00:00:19,200 --> 00:00:21,760
today's show, Andy JA and Adam 
Brewer. 

6
00:00:22,360 --> 00:00:24,480
Welcome to this week's episode 
of the Blue Security. 

7
00:00:24,480 --> 00:00:28,920
Podcast, I'm Andy, your host. 
I'm Adam, your Co host. 

8
00:00:29,840 --> 00:00:32,600
All right, we have 3. 
Topics that we wanted to. 

9
00:00:32,600 --> 00:00:35,960
Talk about today. 
We're going to talk about the. 

10
00:00:36,360 --> 00:00:40,560
Copy fail bug that was. 
Found in Linux, and then we're 

11
00:00:40,560 --> 00:00:42,880
also going to talk. 
About cloud security which? 

12
00:00:42,880 --> 00:00:45,960
Is now in public beta. 
And then finally, at the end of 

13
00:00:45,960 --> 00:00:51,000
the show, we're going to talk 
about how Microsoft is using AI 

14
00:00:51,120 --> 00:00:55,200
in how they defend the company. 
And so since we are going to 

15
00:00:55,200 --> 00:00:58,920
talk about Microsoft, let's just
get that disclaimer right out of

16
00:00:58,920 --> 00:01:01,160
the. 
Way, let's do it. 

17
00:01:01,200 --> 00:01:03,840
Andy and I both work for 
Microsoft in our field security 

18
00:01:03,840 --> 00:01:05,720
sales. 
However, we record the show 

19
00:01:05,720 --> 00:01:09,440
outside of work on our own time.
As I speak we are recording this

20
00:01:09,440 --> 00:01:12,080
at 9:20 AM on a Saturday 
morning. 

21
00:01:12,520 --> 00:01:15,680
So the show is labor of love, 
not a work product of Microsoft.

22
00:01:16,440 --> 00:01:20,520
So with that said, the opinions 
you're about to hear expressed 

23
00:01:20,520 --> 00:01:24,280
are those of Adam Brewer and 
Andy Jaw and do not necessarily 

24
00:01:24,280 --> 00:01:26,440
reflect those of Microsoft 
Corporation. 

25
00:01:26,680 --> 00:01:31,720
And with that, on with the show.
All right, so there was this 

26
00:01:31,840 --> 00:01:36,640
nearly decade old. 
Linux bug. 32 bytes of. 

27
00:01:36,640 --> 00:01:39,960
Python And every major 
distribution is going to be 

28
00:01:39,960 --> 00:01:42,280
vulnerable. 
To this called copy fail, so 

29
00:01:42,280 --> 00:01:43,440
that dropped. 
This week. 

30
00:01:43,440 --> 00:01:46,800
And the big thing is, is that AI
was the one that found it. 

31
00:01:46,800 --> 00:01:50,000
So what is copy fail? 
It is a local privileged 

32
00:01:50,000 --> 00:01:54,840
escalation vulnerability, so 
meaning any unprivileged local 

33
00:01:54,840 --> 00:01:57,520
user can escalate to. 
Root. 

34
00:01:57,760 --> 00:01:59,840
Without needing network. 
Access. 

35
00:02:00,240 --> 00:02:03,760
Or special kernel features and 
it affects every Linux 

36
00:02:03,760 --> 00:02:07,040
distribution that has shipped 
since 2017. 

37
00:02:07,720 --> 00:02:10,080
And unlike most. 
Local. 

38
00:02:10,080 --> 00:02:14,480
Privilege escalations. 
It requires no race window and 

39
00:02:14,480 --> 00:02:16,480
no kernel. 
Specific offsets. 

40
00:02:16,480 --> 00:02:17,600
It's a straight. 
Line logic. 

41
00:02:17,600 --> 00:02:21,960
Flaw that works the. 
Same way every single time and 

42
00:02:21,960 --> 00:02:24,400
the exploit. 
Itself is just. 732 bytes of 

43
00:02:24,400 --> 00:02:28,000
Python using only standard 
library modules and the same 

44
00:02:28,000 --> 00:02:31,480
unmodified script which was 
demonstrated in routing, Ubuntu,

45
00:02:31,840 --> 00:02:35,560
Amazon Linux, RHEL and sussy in 
a single take. 

46
00:02:36,320 --> 00:02:40,320
So the technical breakdowns, if 
you're a Linux person, I know 

47
00:02:40,320 --> 00:02:43,080
like basic Linux. 
So some of the stuff is probably

48
00:02:43,440 --> 00:02:45,920
not like high school. 
On the weeds, isn't it? 

49
00:02:46,000 --> 00:02:48,640
Yeah, but. 
If you are a Linux person and 

50
00:02:48,640 --> 00:02:50,480
you understand this. 
That's great. 

51
00:02:50,480 --> 00:02:52,640
So the bug. 
Lives in Authsense, which is 

52
00:02:53,000 --> 00:02:58,240
auth ENCESNA component of the 
Linux kernel crypto subsystem, 

53
00:02:58,960 --> 00:03:01,640
and it gets chained through the 
AF under score. 

54
00:03:01,640 --> 00:03:03,080
ALG. 
Which is the. 

55
00:03:03,080 --> 00:03:06,840
Kernel, user space, crypto API 
and splice. 

56
00:03:06,840 --> 00:03:10,640
To achieve A4. 
Byte write into the page cache. 

57
00:03:10,880 --> 00:03:13,000
The Four. 
Byte write is enough to corrupt.

58
00:03:13,000 --> 00:03:17,200
The suit. 
ID binary which is the SU in 

59
00:03:17,200 --> 00:03:20,640
memory, not on disk, so it won't
survive. 

60
00:03:20,640 --> 00:03:24,720
A reboot, but it does. 
Hand you a real root shell in 

61
00:03:24,720 --> 00:03:31,280
the meantime, and so the AF 
under score ALG ships enabled in

62
00:03:31,280 --> 00:03:36,560
essentially every mainstream 
distro's default configuration, 

63
00:03:36,560 --> 00:03:42,120
which is why the entire window 
from 2017 to patch is in. 

64
00:03:42,120 --> 00:03:47,320
Scope out-of-the-box. 
So who's at risk, really? 

65
00:03:47,800 --> 00:03:51,800
Any multi tenant Linux host that
share unlike a shared dev. 

66
00:03:51,800 --> 00:03:55,320
Box or. 
Jump, host, or build servers are

67
00:03:55,320 --> 00:03:58,960
all at risk. 
Any user on those systems can 

68
00:03:58,960 --> 00:04:02,360
become root. 
Kubernetes clusters are 

69
00:04:02,360 --> 00:04:05,560
particularly dangerous. 
Because the page cache is shared

70
00:04:05,560 --> 00:04:08,400
across the host. 
Meaning a compromise pod can 

71
00:04:08,400 --> 00:04:12,040
then cross tenant boundaries and
take over the entire node. 

72
00:04:13,080 --> 00:04:16,000
That's bad. 
Yeah, that's bad. 

73
00:04:16,000 --> 00:04:20,079
CI runners, GitHub action self 
host runners, Git lab runners, 

74
00:04:20,079 --> 00:04:22,880
Jenkins agents are. 
Also high risk because. 

75
00:04:23,280 --> 00:04:26,920
They execute untrusted. 
PR code as regular users. 

76
00:04:26,920 --> 00:04:31,040
On a shared kernel, meaning a 
malicious pull request could 

77
00:04:31,640 --> 00:04:35,640
become root on the runner. 
So you know, one of the things 

78
00:04:35,640 --> 00:04:37,960
that was interesting as I. 
Read through this was that? 

79
00:04:37,960 --> 00:04:41,320
It was discovered by an AI 
scanner, so. 

80
00:04:41,320 --> 00:04:45,880
Copy fail was discovered by. 
Zint code that's spelled XINT 

81
00:04:46,720 --> 00:04:50,080
and it's an AI powered 
vulnerability scanner after 

82
00:04:50,080 --> 00:04:53,160
about an hour of scan time, and 
it pointed at a Linux crypto 

83
00:04:53,160 --> 00:04:57,760
subsystem with no custom harness
and just one operator prompt. 

84
00:04:58,320 --> 00:05:00,880
This is a. 
Significant discovery because 

85
00:05:00,880 --> 00:05:02,880
traditional. 
Security would have required. 

86
00:05:02,960 --> 00:05:06,120
You know, like a skilled human 
researcher spending days or 

87
00:05:06,120 --> 00:05:09,440
weeks manually auditing current 
code to find a bug like. 

88
00:05:09,440 --> 00:05:11,560
This. 
And the same scan also. 

89
00:05:11,560 --> 00:05:14,400
Surfaced other. 
High severity bugs that are 

90
00:05:14,400 --> 00:05:18,040
still in coordinated disclosure.
So this is just the. 

91
00:05:18,040 --> 00:05:20,600
First one out the door, but 
there's more to come. 

92
00:05:20,800 --> 00:05:24,560
And so, you know, we've been 
talking about AI in the last 

93
00:05:24,560 --> 00:05:28,200
couple of shows and this is just
another example that really 

94
00:05:28,200 --> 00:05:31,040
shows that vulnerability 
scanning, especially AI powered 

95
00:05:31,040 --> 00:05:34,080
one, is no longer theoretical. 
It's finding production 

96
00:05:34,160 --> 00:05:39,040
exploitable, decades old bugs 
and critical infrastructure. 

97
00:05:39,960 --> 00:05:44,360
So no, like, you know, real 
discovery here, the primary fix 

98
00:05:44,360 --> 00:05:47,960
is really just patching. 
So look for a main mainline 

99
00:05:47,960 --> 00:05:49,640
commit. 
Which will revert. 

100
00:05:49,640 --> 00:05:53,880
The 2017 optimization that was 
induced introduced with this 

101
00:05:53,880 --> 00:05:57,800
flaw, and if you can't patch 
immediately, you can disable the

102
00:05:57,840 --> 00:06:02,360
AL GIF under score amid module 
as a stop gap. 

103
00:06:02,360 --> 00:06:05,080
For most systems, this won't 
break any measurable anything 

104
00:06:05,080 --> 00:06:08,000
that's measurable. 
Because there are very few. 

105
00:06:08,000 --> 00:06:11,720
Applications that actually call 
the AF under score ALG. 

106
00:06:11,720 --> 00:06:14,720
Directly. 
So for untrusted workloads like 

107
00:06:14,720 --> 00:06:17,920
containers and CI runners, you 
can also block the AF under 

108
00:06:17,920 --> 00:06:22,600
SCORE ALG socket creation via 
seccomp, and that's recommended 

109
00:06:22,600 --> 00:06:24,880
regardless. 
Of the patch status. 

110
00:06:24,920 --> 00:06:26,880
As well. 
Really interesting stuff. 

111
00:06:26,920 --> 00:06:30,240
If you're a Linux nerd, this 
probably all makes sense to you.

112
00:06:30,240 --> 00:06:33,600
To me, I'm just like. 
OK, this is really bad. 

113
00:06:34,160 --> 00:06:37,360
I got to look at all my Linux 
distros and look for a patch and

114
00:06:37,360 --> 00:06:39,280
get that done as soon as 
possible. 

115
00:06:39,560 --> 00:06:43,720
We talked about AI vulnerability
scanning and AI finding new 

116
00:06:43,720 --> 00:06:47,400
vulnerabilities, and one of the 
specific things we pointed out 

117
00:06:47,960 --> 00:06:51,120
is that AI is far better at 
finding chained vulnerabilities 

118
00:06:51,120 --> 00:06:53,680
than humans are because it's so 
iterative. 

119
00:06:53,720 --> 00:06:57,600
Humans lack, I mean, we we need 
to sleep, we need to eat, we 

120
00:06:57,600 --> 00:07:00,640
need to do other things, we need
to stay interested very 

121
00:07:00,640 --> 00:07:03,840
systematically testing different
chaining combinations for 

122
00:07:03,840 --> 00:07:08,000
weaknesses as such a strength of
machines. 

123
00:07:08,000 --> 00:07:10,880
And so this is a chained 
vulnerability. 

124
00:07:10,880 --> 00:07:13,680
To be clear, you kind of walk 
through it when we first started

125
00:07:13,680 --> 00:07:16,640
talking about it, Andy. 
So it's a great example of that 

126
00:07:16,640 --> 00:07:21,320
risk. 
And so ultimately, I, I think 

127
00:07:21,320 --> 00:07:23,640
the reaction here just needs to 
be measured and and this is 

128
00:07:23,640 --> 00:07:27,160
where we just say, OK, you know,
this is no longer a flying car. 

129
00:07:27,160 --> 00:07:31,600
This is a real thing. 
And thankfully, the way we 

130
00:07:31,640 --> 00:07:34,800
handle this is the same way we 
handle anything else that's been

131
00:07:34,800 --> 00:07:37,480
disclosed through coordinated 
responsible vulnerability. 

132
00:07:37,480 --> 00:07:41,560
We patch in a timely fashion, 
but we don't scream the sky is 

133
00:07:41,560 --> 00:07:43,720
falling. 
We we patch, we move forward, 

134
00:07:43,720 --> 00:07:46,680
and we await the next one. 
And apparently more are coming. 

135
00:07:46,680 --> 00:07:50,440
And ultimately, if the rate of 
change becomes unsustainable, we

136
00:07:50,440 --> 00:07:52,320
will develop alternatives at 
that time. 

137
00:07:52,320 --> 00:07:55,360
The one thing is necessity is 
the mother of all invention. 

138
00:07:55,360 --> 00:08:00,640
And so for right now we patch 
our Linux based systems and we 

139
00:08:00,640 --> 00:08:03,240
move forward. 
And This is why we're being 

140
00:08:03,320 --> 00:08:06,480
conscientious about this and 
about the deployment and 

141
00:08:06,480 --> 00:08:09,920
disbursement of new models like 
Mythos that will find these and,

142
00:08:09,920 --> 00:08:13,880
and giving the Linux Foundation 
and other very major critical 

143
00:08:13,880 --> 00:08:17,240
technology providers are the 
opportunity to get out in front 

144
00:08:17,240 --> 00:08:19,160
of this. 
And I know we'll talk a little 

145
00:08:19,160 --> 00:08:21,440
bit more about what Microsoft's 
doing in this space as well 

146
00:08:21,680 --> 00:08:23,760
later on. 
Our employer, who I think brings

147
00:08:23,760 --> 00:08:26,640
an interesting perspective to 
the table as I mean, I think 

148
00:08:26,640 --> 00:08:28,800
it's pretty fair to say 
Microsoft has issued more 

149
00:08:28,800 --> 00:08:31,200
security patches than any other 
security vendor in the world. 

150
00:08:31,520 --> 00:08:34,679
And it, and it's gotten it down 
to a very sustainable, 

151
00:08:34,679 --> 00:08:39,039
predictable cadence and, and 
patching 1.6 billion Windows 

152
00:08:39,039 --> 00:08:41,520
endpoints around the world every
single month on the second 

153
00:08:41,520 --> 00:08:45,840
Tuesday of the month without 
really any notable disruption or

154
00:08:45,840 --> 00:08:47,920
impact or anything else is, is 
pretty impressive. 

155
00:08:47,920 --> 00:08:51,840
So you know, the Apple does a 
great job patching billions of 

156
00:08:51,840 --> 00:08:55,240
iOS, Mac OS derivative devices 
as well. 

157
00:08:55,240 --> 00:08:57,360
And, and obviously Google does a
great job. 

158
00:08:57,800 --> 00:09:00,640
They weren't at first I was 
highly critical of them, but now

159
00:09:00,640 --> 00:09:02,720
patching billions of Android 
endpoints around the world. 

160
00:09:02,720 --> 00:09:05,960
So it's, it's something we've 
gotten a lot better at and do a 

161
00:09:05,960 --> 00:09:08,760
lot better at scale. 
So you know, fingers crossed 

162
00:09:08,760 --> 00:09:12,000
that all that muscle we have 
built for many, many years just 

163
00:09:12,000 --> 00:09:14,800
carries forward and we just kind
of keep doing the same thing. 

164
00:09:14,800 --> 00:09:17,360
And you know, if we have 0 days,
we'll address those. 

165
00:09:17,600 --> 00:09:20,640
Otherwise, if they're kind of 
routine coordinated responsible 

166
00:09:20,640 --> 00:09:23,800
disclosure, we'll patch them on 
our predictable cadence and 

167
00:09:23,800 --> 00:09:26,920
we'll move on. 
So interesting, but kind of not 

168
00:09:26,920 --> 00:09:28,440
interesting at the same time, 
right. 

169
00:09:28,680 --> 00:09:31,600
So in terms of interesting, in 
terms of it's novel and new and 

170
00:09:31,600 --> 00:09:35,240
different, but uninteresting in 
terms of our response, it is the

171
00:09:35,240 --> 00:09:37,720
same way we would respond to any
other vulnerability discovered 

172
00:09:37,720 --> 00:09:40,400
any other way and onward and 
upward. 

173
00:09:40,440 --> 00:09:45,160
So let's keep going. 
So Cloud security is now in 

174
00:09:45,240 --> 00:09:47,640
public beta. 
And so anthropic. 

175
00:09:47,640 --> 00:09:49,560
Just shipped A dedicated 
security. 

176
00:09:49,560 --> 00:09:53,120
Product and it is different than
what we had, you know, talked. 

177
00:09:53,120 --> 00:09:55,440
About a few weeks. 
Ago, which was Mythos, which was

178
00:09:55,440 --> 00:09:59,360
that kind of frontier model that
they're not releasing to the 

179
00:09:59,360 --> 00:10:02,120
general public, but this one is 
and so. 

180
00:10:02,400 --> 00:10:06,000
Cloud Security is a code. 
Vulnerability scanner that's. 

181
00:10:06,000 --> 00:10:09,800
Powered by Cloud Opus 4. .7 
which is kind of like their most

182
00:10:09,800 --> 00:10:13,040
powerful language that's out in 
production day. 

183
00:10:13,640 --> 00:10:15,400
It's available directly through 
cloud. 

184
00:10:15,480 --> 00:10:18,920
AI sidebar or at cloud? 
Dot AI security. 

185
00:10:18,920 --> 00:10:22,560
It doesn't require any API 
integrations or custom agent 

186
00:10:22,560 --> 00:10:25,520
build. 
It's in public beta for Cloud 

187
00:10:25,520 --> 00:10:29,480
enterprise customers, with 
access for team and Max 

188
00:10:29,480 --> 00:10:33,560
customers coming soon. 
It was previously known as Cloud

189
00:10:33,560 --> 00:10:36,720
Code Security and already has 
been tested by hundreds of 

190
00:10:36,720 --> 00:10:39,720
organizations during a limited. 
Research preview. 

191
00:10:40,120 --> 00:10:41,960
And so all of their feedback 
kind of. 

192
00:10:41,960 --> 00:10:45,280
Shaped this specific release. 
Pretty simple to use all you 

193
00:10:45,280 --> 00:10:46,920
need to do. 
Is select a repository. 

194
00:10:47,360 --> 00:10:49,840
Scope it down to. 
A specific directory or. 

195
00:10:49,840 --> 00:10:53,640
Branch and then you can kick off
a scan rather than matching 

196
00:10:53,640 --> 00:10:58,080
known vulnerability patterns. 
Clawed reasons over the code. 

197
00:10:58,280 --> 00:11:00,040
Like a security researcher 
would. 

198
00:11:00,360 --> 00:11:02,800
And it understands how 
components interact across 

199
00:11:02,800 --> 00:11:06,520
files, modules, tracing data 
flows and reading the source. 

200
00:11:06,520 --> 00:11:08,960
To build up a picture. 
Of what could go wrong. 

201
00:11:09,520 --> 00:11:13,360
Each finding comes with a 
confidence rating, a severity 

202
00:11:13,360 --> 00:11:17,240
assessment, likely impact 
reproduction steps, and 

203
00:11:17,240 --> 00:11:22,160
generated patch that you can 
openly direct directly in cloud 

204
00:11:22,160 --> 00:11:24,000
code. 
To work through a fix. 

205
00:11:24,000 --> 00:11:28,720
In the context, you can also 
schedule scans on a regular 

206
00:11:28,720 --> 00:11:31,120
cadence and scope. 
It to a specific directory. 

207
00:11:31,320 --> 00:11:33,920
Directory within a repo. 
You can dismiss findings with 

208
00:11:33,920 --> 00:11:36,760
documented reasons so future 
reviewers know. 

209
00:11:37,120 --> 00:11:39,680
That prior triage? 
Was intentional and you can 

210
00:11:39,680 --> 00:11:43,560
export findings as CSVS or 
markdowns for existing tracking 

211
00:11:43,560 --> 00:11:46,080
systems. 
There are also webhooks and then

212
00:11:46,080 --> 00:11:48,760
you can pipe scan directly into 
Slack. 

213
00:11:48,800 --> 00:11:52,400
JIRA or other tools? 
That your security team already 

214
00:11:52,400 --> 00:11:54,120
lives in. 
So that closes the gap. 

215
00:11:54,120 --> 00:11:58,840
Between your finding the 
discovery and then having that 

216
00:11:58,840 --> 00:12:01,000
engineer. 
Aware of it so. 

217
00:12:01,000 --> 00:12:02,920
You know, like I said in the 
beginning here. 

218
00:12:02,920 --> 00:12:06,440
It is a separate. 
Kind of tool. 

219
00:12:06,440 --> 00:12:10,440
Then the Cloud Mythos preview 
which is a more powerful model. 

220
00:12:10,720 --> 00:12:14,200
That is supposed to. 
Surpass like the most elite 

221
00:12:14,200 --> 00:12:17,920
human experts, that one right 
now is only being released to 

222
00:12:18,000 --> 00:12:20,880
some vetted partners to that 
project last ring that we 

223
00:12:20,880 --> 00:12:23,240
talked. 
About whereas cloud security 

224
00:12:23,240 --> 00:12:25,640
here is broadly available, it's.
An enterprise grade. 

225
00:12:25,640 --> 00:12:28,320
Product it puts Opus. .7's 
capability. 

226
00:12:28,320 --> 00:12:31,320
To work across your code base. 
For practical vulnerability 

227
00:12:31,320 --> 00:12:35,720
scanning patch generation 
without requiring any kind of 

228
00:12:35,720 --> 00:12:39,840
control controlled access that 
Mythos normally has in the 

229
00:12:39,840 --> 00:12:43,160
partner ecosystem. 
Crowdstrike, Microsoft Security,

230
00:12:43,160 --> 00:12:46,000
Palo Alto Networks, Sentinel, 
One Trend AI Wiz. 

231
00:12:46,000 --> 00:12:49,360
They're all embedding Opus 4.7 
into their existing security. 

232
00:12:49,360 --> 00:12:52,280
Tools. 
So organisations can access 

233
00:12:52,280 --> 00:12:55,120
these capabilities through the 
platforms they already run and 

234
00:12:55,120 --> 00:12:58,920
then on the services side like 
Accenture, Deloitte, PwC, 

235
00:12:58,920 --> 00:13:01,800
they're all helping enterprises 
deploy cloud integrated security

236
00:13:01,800 --> 00:13:03,480
solutions for vulnerability 
management. 

237
00:13:03,920 --> 00:13:06,520
Secured code review. 
And incident response. 

238
00:13:06,800 --> 00:13:09,360
I think this is great. 
You know, if you're already 

239
00:13:09,360 --> 00:13:13,880
using Cloud code, I think this 
integrates, you know, seamlessly

240
00:13:13,880 --> 00:13:17,200
into that workflow. 
I'm really curious though how 

241
00:13:17,200 --> 00:13:21,440
this will compete with other 
code scanning like tools that 

242
00:13:21,440 --> 00:13:25,720
have been traditionally if you 
have an apps app SEC program and

243
00:13:25,720 --> 00:13:28,320
you're already using code 
scanning like something like I 

244
00:13:28,320 --> 00:13:31,160
used a product like Veracode in 
the past, there's GitHub 

245
00:13:31,160 --> 00:13:33,920
Advanced Security and you know 
how those are going to be 

246
00:13:33,920 --> 00:13:38,120
integrated into maybe either use
this or how this is going to 

247
00:13:38,120 --> 00:13:39,800
compete. 
With stuff like that, so. 

248
00:13:41,000 --> 00:13:44,560
I think this is anything that 
increases the security of code. 

249
00:13:44,560 --> 00:13:45,640
I think is good. 
I'm just. 

250
00:13:45,640 --> 00:13:48,000
Curious how this will shake. 
Out in the security kind of 

251
00:13:48,400 --> 00:13:51,120
solutions industry because it is
something that you have. 

252
00:13:51,120 --> 00:13:53,240
To pay for with. 
Cloud Enterprise. 

253
00:13:53,640 --> 00:13:59,120
Yeah, and Opus 4.7 is more than 
is a general purpose deep 

254
00:13:59,120 --> 00:14:02,880
thinking reasoning model. 
So it's it's used for all those 

255
00:14:02,880 --> 00:14:05,800
scenarios. 
In fact, in in our world, in the

256
00:14:05,800 --> 00:14:10,240
Microsoft 365 Copilot Researcher
Agent Opus 4.7 is now available 

257
00:14:10,240 --> 00:14:14,600
as a model you can use for for 
doing research Funny. 

258
00:14:14,600 --> 00:14:16,320
Enough. 
You mentioned that I used it 

259
00:14:16,320 --> 00:14:19,160
recently this week and what it 
did. 

260
00:14:19,160 --> 00:14:21,080
Was. 
It used. 

261
00:14:21,080 --> 00:14:23,160
ChatGPT. 
The ChatGPT. 

262
00:14:23,160 --> 00:14:28,440
Model to 1st generate an output 
and then it used Opus to then 

263
00:14:28,480 --> 00:14:30,840
refine it. 
So it used multiple models 

264
00:14:30,840 --> 00:14:33,960
within the researcher agent. 
Which I thought was pretty cool.

265
00:14:34,320 --> 00:14:37,000
Yeah, that that feature recently
shipped. 

266
00:14:37,120 --> 00:14:40,480
It's like effectively peer 
review where it runs it with the

267
00:14:40,480 --> 00:14:43,720
GPT model 5.5 now and then it 
will review it with the 

268
00:14:43,720 --> 00:14:46,520
Anthropic model. 
Or you can also do what's called

269
00:14:46,520 --> 00:14:50,240
Model Council, which sounds like
tribal council to me, But it'll 

270
00:14:50,240 --> 00:14:54,280
actually it'll run the query 
across both like an anthropic 

271
00:14:54,280 --> 00:14:58,720
and open AI model at the same 
time and then give you like dual

272
00:14:58,720 --> 00:15:01,760
output side by side and you can 
compare like how they both did. 

273
00:15:01,760 --> 00:15:05,160
So if you, if you want to do 
that, it's a really clean way to

274
00:15:05,160 --> 00:15:07,920
get multiple outputs and see 
which one you like better and 

275
00:15:07,920 --> 00:15:10,800
which ones better tailored do 
what you were aiming for. 

276
00:15:11,560 --> 00:15:15,000
So that's, you know, neither 
here nor there, but just an 

277
00:15:15,000 --> 00:15:17,640
interesting side note that I 
want to point out. 

278
00:15:17,640 --> 00:15:21,440
Opus 4.7 is a general purpose, 
you know, deep thinking 

279
00:15:21,480 --> 00:15:26,560
reasoning model and it's being 
used in this scenario to analyze

280
00:15:26,560 --> 00:15:31,040
code and and Mythos is we're 
also comparing it to will never 

281
00:15:31,040 --> 00:15:32,920
ship. 
They have said that Mythos is 

282
00:15:32,920 --> 00:15:36,320
not a product that's designed to
move to production. 

283
00:15:36,320 --> 00:15:39,520
It's it's designed to be a 
research preview to help inform 

284
00:15:39,600 --> 00:15:42,600
production releases. 
And apparently Opus 4.7 is 

285
00:15:42,600 --> 00:15:47,800
pretty close already to Mythos 
capability set and expect that I

286
00:15:47,920 --> 00:15:51,040
I would expect more iterative 
updates to Opus and, and 

287
00:15:51,040 --> 00:15:53,120
eventually they'll say, yeah, we
are now at Mythos level 

288
00:15:53,120 --> 00:15:55,440
capability is kind of how this 
will proceed. 

289
00:15:55,800 --> 00:15:57,600
That's just me reading the tea 
leaves, but that's my 

290
00:15:57,600 --> 00:15:59,400
prediction. 
You know, this is specifically 

291
00:15:59,400 --> 00:16:02,680
code scanning, not necessarily 
him at a point you at a system 

292
00:16:02,680 --> 00:16:05,200
where I don't have access to the
code code and I want you to 

293
00:16:05,200 --> 00:16:08,240
reason through how you would 
attack this or or anything else.

294
00:16:08,320 --> 00:16:11,400
Obviously code scanning is 
highly important right now and 

295
00:16:11,400 --> 00:16:14,520
that's a lot of reason why 
organizations got access to 

296
00:16:14,520 --> 00:16:18,240
Mythos, especially companies 
that are not open source like 

297
00:16:18,240 --> 00:16:20,040
our employer. 
Run it against the Windows code 

298
00:16:20,040 --> 00:16:22,280
base, see what you come up with,
and then we can start 

299
00:16:22,280 --> 00:16:27,200
proactively fixing stuff. 
Not to say I, I know or can 

300
00:16:27,200 --> 00:16:30,040
confirm or deny that's been done
that I'm not, you know, read 

301
00:16:30,040 --> 00:16:32,400
into that, but I assume 
something like that is happening

302
00:16:32,440 --> 00:16:34,600
anyhow. 
So this is this is cool and 

303
00:16:34,600 --> 00:16:36,000
we'll continue to see this 
develop. 

304
00:16:36,000 --> 00:16:39,200
And This is why I think I said 
either last week or the week 

305
00:16:39,200 --> 00:16:43,480
before, I continue to be bullish
that in the long run, this AI 

306
00:16:43,480 --> 00:16:49,040
will actually be a great 
equalizer for the good guys, for

307
00:16:49,040 --> 00:16:51,960
the blue hats of the world, for 
our blue security podcast 

308
00:16:51,960 --> 00:16:56,600
listeners, because it gives us 
as it gives us basically that 

309
00:16:56,600 --> 00:16:59,480
equalizing capability through 
effectively money. 

310
00:16:59,960 --> 00:17:03,200
We all are our security teams 
are reasonably well funded and 

311
00:17:03,200 --> 00:17:05,480
gives us the chance to kind of 
get out in front of these. 

312
00:17:05,480 --> 00:17:09,960
Even if we may not have 
headcount or long term advantage

313
00:17:09,960 --> 00:17:13,240
from a human capital 
perspective, we can gain an 

314
00:17:13,240 --> 00:17:16,880
advantage. 
That's that's consistent and 

315
00:17:16,960 --> 00:17:20,280
persistent and durable through 
the use of these tools in the 

316
00:17:20,280 --> 00:17:22,720
long run. 
And I I think it will actually 

317
00:17:22,720 --> 00:17:25,960
be difficult and I'm optimistic 
for the bad guys to keep up 

318
00:17:25,960 --> 00:17:27,400
here. 
But in the short term, yes, 

319
00:17:27,400 --> 00:17:30,000
there's going to be pain as we 
first transition to this world 

320
00:17:30,000 --> 00:17:32,920
and we have a flurry of new 
vulnerabilities announced all at

321
00:17:32,920 --> 00:17:35,040
once. 
Yeah, It may be difficult for 

322
00:17:35,040 --> 00:17:37,760
the next couple of years, but 
then we will reach a steady 

323
00:17:37,760 --> 00:17:41,600
state where all major code bases
have been reviewed, all the 

324
00:17:41,600 --> 00:17:45,600
major, the quote UN quote low 
hanging fruit has been found and

325
00:17:45,600 --> 00:17:47,880
has been patched. 
And at that point moving 

326
00:17:47,880 --> 00:17:51,680
forward, if we're running these 
types of vulnerability scanners 

327
00:17:51,680 --> 00:17:55,720
against new code before it ever 
ships, we are shipping defect 

328
00:17:55,720 --> 00:17:58,800
free, security free code from 
the get go. 

329
00:17:59,600 --> 00:18:02,120
And then there's just less 
opportunity or less need to go 

330
00:18:02,120 --> 00:18:05,880
back and patch it later. 
So again, short term pain here. 

331
00:18:06,160 --> 00:18:09,160
The sky is not falling, although
things might suck for a little 

332
00:18:09,160 --> 00:18:11,960
bit. 
But in the long run, the goal is

333
00:18:11,960 --> 00:18:15,440
that code just leaves, leaves 
the door and it's secure by 

334
00:18:15,440 --> 00:18:19,440
design and secure by default. 
And there's a lot less we have 

335
00:18:19,440 --> 00:18:22,160
to do as defenders to run around
and patched up after the fact. 

336
00:18:22,560 --> 00:18:25,240
And that will be a great day. 
It's not to say we'll never do 

337
00:18:25,240 --> 00:18:28,440
it again, but I bet it'll become
less frequent because I think a 

338
00:18:28,440 --> 00:18:31,320
lot more code is going to ship 
in a lot better state with a lot

339
00:18:31,320 --> 00:18:35,000
more security review from before
it ever ships. 

340
00:18:35,000 --> 00:18:38,600
So that's why tools like this 
I'm very optimistic about in the

341
00:18:38,600 --> 00:18:40,800
long run, even if there is some 
short term pain. 

342
00:18:41,480 --> 00:18:43,960
Yeah, I did see. 
Something like or. 

343
00:18:43,960 --> 00:18:46,800
I had a colleague mentioned 
earlier this week that 

344
00:18:47,720 --> 00:18:51,400
essentially all the information 
that has been generated has 

345
00:18:51,400 --> 00:18:54,840
already been reasoned over with 
the current AI models. 

346
00:18:54,840 --> 00:18:57,000
Like there's nothing that has 
been generated, like all the 

347
00:18:57,000 --> 00:19:01,240
information of the Internet has 
already been, you know, sucked 

348
00:19:01,240 --> 00:19:03,880
into these models and reasoned. 
Over so so we're. 

349
00:19:04,160 --> 00:19:07,800
At a point where you know it's 
only new information or it's 

350
00:19:07,800 --> 00:19:10,960
reasoning over like existing 
information and on itself. 

351
00:19:11,280 --> 00:19:14,040
So like I think it like you're 
saying at some point all these 

352
00:19:14,040 --> 00:19:16,240
discoveries will eventually. 
Be less. 

353
00:19:16,240 --> 00:19:19,160
Frequent we're going to have 
everything patched up and then 

354
00:19:19,160 --> 00:19:21,320
only new stuff will. 
Will exist. 

355
00:19:21,480 --> 00:19:23,640
Yeah, you get to a steady state.
Here we are. 

356
00:19:23,640 --> 00:19:27,280
We are in a inflection point 
right now where there is a rapid

357
00:19:27,280 --> 00:19:30,880
rate of change and sure, 
publicly facing information on 

358
00:19:30,880 --> 00:19:33,440
the Internet has all been 
ingested and incorporated into 

359
00:19:33,440 --> 00:19:36,040
models. 
Private code bases, especially 

360
00:19:36,040 --> 00:19:39,360
massive private code bases, may 
not have been fully reasoned 

361
00:19:39,360 --> 00:19:44,000
over yet. 
Maybe partially, maybe in in 

362
00:19:44,000 --> 00:19:48,080
wide swaths, but not all of it. 
And then as models continue to 

363
00:19:48,080 --> 00:19:50,880
evolve the state-of-the-art and 
they get better at reasoning and

364
00:19:50,880 --> 00:19:53,520
thinking and discovering 
vulnerabilities, even though 

365
00:19:53,520 --> 00:19:56,080
they've seen the code base 
before, they can take a fresh 

366
00:19:56,080 --> 00:19:59,360
look at it, a fresh swag at it 
and do a better job discovering.

367
00:19:59,360 --> 00:20:02,680
So I think we're saying the same
thing that the steady state is 

368
00:20:02,680 --> 00:20:05,080
coming, but we are not in the 
steady state right now. 

369
00:20:05,080 --> 00:20:07,400
We are getting there. 
We are, we are climbing the 

370
00:20:07,400 --> 00:20:10,040
difficult side of the mountain 
right now, but we will reach 

371
00:20:10,040 --> 00:20:13,040
that summit, we will level off 
at some point. 

372
00:20:13,400 --> 00:20:17,520
But right now is we are in the 
hard times right now before we 

373
00:20:17,600 --> 00:20:20,240
get to the summit and it we're 
coming down the other side of 

374
00:20:20,240 --> 00:20:21,960
the mountain and it gets a lot 
easier. 

375
00:20:22,280 --> 00:20:24,320
But this is the hard part right 
now. 

376
00:20:24,320 --> 00:20:27,080
So don't lose faith, don't lose 
hope. 

377
00:20:28,040 --> 00:20:30,000
I get it. 
This is, this is going to be, 

378
00:20:30,000 --> 00:20:31,120
there's going to be a lot of 
patching. 

379
00:20:31,400 --> 00:20:33,560
There just is, there's going to 
be a lot of 0 days. 

380
00:20:33,560 --> 00:20:36,480
There's going to be a lot of 
high level, high severity 

381
00:20:36,480 --> 00:20:38,960
vulnerabilities that we have to 
address very rapidly. 

382
00:20:38,960 --> 00:20:42,040
Lot of fire drills. 
Let's just prepare for it. 

383
00:20:42,360 --> 00:20:45,680
Let's get really crisp on, on 
how we patch and we should 

384
00:20:45,680 --> 00:20:48,200
already be, although it, we're 
always getting better. 

385
00:20:48,200 --> 00:20:51,200
We, we have that growth mindset,
the continuous improvement 

386
00:20:51,200 --> 00:20:54,080
mindset and we'll continue to 
get better at it. 

387
00:20:54,080 --> 00:20:58,440
But I do think it is not hard to
imagine a day three years from 

388
00:20:58,440 --> 00:21:02,000
now, Andy before the turn of the
decade where where we patch a 

389
00:21:02,000 --> 00:21:05,480
lot less frequently and patching
is a lot less of security 

390
00:21:05,480 --> 00:21:08,880
strategy. 
I, I do think that day comes and

391
00:21:08,880 --> 00:21:13,280
it shifts back to probably 
picking off some of these last 

392
00:21:13,280 --> 00:21:16,560
like human based changes that 
organizations have been hesitant

393
00:21:16,560 --> 00:21:19,360
to bite off. 
Things like getting back to fish

394
00:21:19,360 --> 00:21:21,640
resistant MFA and getting that 
fully deployed. 

395
00:21:21,640 --> 00:21:24,320
And those are challenges we 
haven't bit off because they're 

396
00:21:24,320 --> 00:21:26,000
not really technology 
challenges. 

397
00:21:26,000 --> 00:21:28,760
The technology is solved. 
It's a human challenge. 

398
00:21:29,080 --> 00:21:30,760
And so when do you have 
management buy in? 

399
00:21:30,760 --> 00:21:33,560
When do you have employee based 
buy in to make those changes and

400
00:21:33,560 --> 00:21:35,760
implement them? 
We may, we may have to shift our

401
00:21:35,760 --> 00:21:39,280
focus temporarily as we deal 
with a lot of these discovered 

402
00:21:39,280 --> 00:21:42,360
vulnerabilities. 
And then when we're doing less 

403
00:21:42,360 --> 00:21:45,720
patching, then we'll have more 
time theoretically to focus on 

404
00:21:46,000 --> 00:21:48,360
again those human impacting 
changes that we still need to 

405
00:21:48,360 --> 00:21:52,560
make and still need to get done.
I am not, by the way, I'm not 

406
00:21:52,560 --> 00:21:55,200
encouraging you to delay your 
fish resistant MFA deployment 

407
00:21:55,200 --> 00:21:57,400
for three more years. 
You should go do it tomorrow. 

408
00:21:57,680 --> 00:22:00,040
But if you don't have management
by and if you don't have the 

409
00:22:00,040 --> 00:22:02,480
headcount, if you don't have the
bandwidth, I get it. 

410
00:22:02,840 --> 00:22:06,960
But I just want to be clear, I'm
not advocating for that, but I 

411
00:22:06,960 --> 00:22:09,080
get it. 
Some orgs, they're going to be 

412
00:22:09,080 --> 00:22:12,360
so heads down with some of these
patching exercises right now. 

413
00:22:12,360 --> 00:22:14,320
Some of this work may get 
delayed that they've been 

414
00:22:14,320 --> 00:22:17,440
meaning to do. 
All right, our final topic, I 

415
00:22:17,440 --> 00:22:20,000
wanted to talk about Microsoft's
AI security. 

416
00:22:20,000 --> 00:22:23,640
Strategy because we published. 
A full road map and it's a multi

417
00:22:23,640 --> 00:22:26,560
layered security strategy and 
not really like a. 

418
00:22:26,560 --> 00:22:28,360
Specific. 
Product announcement or anything

419
00:22:28,360 --> 00:22:31,280
like that, It's a. 
Little product announcement 

420
00:22:31,280 --> 00:22:35,200
sprinkled in there like a little
tease, but yes, overall this was

421
00:22:35,200 --> 00:22:37,920
much more. 
This was on the SRC blog so the 

422
00:22:37,920 --> 00:22:41,080
Microsoft off security. 
Research Center. 

423
00:22:41,360 --> 00:22:42,760
Research Center, Thank you. 
Yes. 

424
00:22:42,760 --> 00:22:47,800
So I mean explicitly not a 
product group, not a revenue 

425
00:22:47,800 --> 00:22:52,680
generating part of the company, 
a A-Team who has a relatively 

426
00:22:52,680 --> 00:22:55,840
altruistic purpose of protect 
our world, protect Microsoft, 

427
00:22:55,840 --> 00:22:59,000
protect our world, protect our 
customers, that that is their 

428
00:22:59,000 --> 00:23:01,520
focus. 
So this is very explicitly not 

429
00:23:01,920 --> 00:23:04,200
designed to make money or 
generate revenue. 

430
00:23:04,200 --> 00:23:06,720
It's designed to keep people 
safe and articulate how we're 

431
00:23:06,720 --> 00:23:08,000
going to do that. 
Yeah. 

432
00:23:08,200 --> 00:23:12,040
And so at the front of the. 
Blog, I should say. 

433
00:23:13,000 --> 00:23:17,440
They talk about how recent AI 
advances now mean that models 

434
00:23:17,440 --> 00:23:19,600
can autonomously discover 
weaknesses. 

435
00:23:19,600 --> 00:23:21,720
They can chain multiple. 
Lower severity issues. 

436
00:23:21,720 --> 00:23:24,000
Into working end to end 
exploits, things that we've 

437
00:23:24,000 --> 00:23:28,840
talked about on our show before.
And now you're essentially AI is

438
00:23:28,840 --> 00:23:31,840
significantly compressing that 
window from vulnerability 

439
00:23:31,840 --> 00:23:35,160
discovery to exploitation. 
And so Microsoft's argument is 

440
00:23:35,160 --> 00:23:37,720
that these same capabilities 
that are giving attackers an 

441
00:23:37,720 --> 00:23:41,240
edge can also create a genuine 
opportunity for defenders. 

442
00:23:41,640 --> 00:23:45,760
It's almost like we are talking 
about the blog in the previous 

443
00:23:45,760 --> 00:23:48,360
few minutes of this show, like 
Adam, you were making the same 

444
00:23:48,360 --> 00:23:50,840
point, right? 
And so Hive. 

445
00:23:50,840 --> 00:23:54,960
Mind is well at work. 
That, you know over the past two

446
00:23:54,960 --> 00:23:58,680
years Microsoft has had this 
secure future initiative and 

447
00:23:58,680 --> 00:24:01,600
through that they've been 
strengthening their security 

448
00:24:01,600 --> 00:24:04,240
foundations through AI to 
accelerate vulnerability 

449
00:24:04,240 --> 00:24:07,760
discovery and remediation and 
now there's even more so in. 

450
00:24:07,760 --> 00:24:11,240
This they published the three 
pillars of the strategy. 

451
00:24:11,240 --> 00:24:13,360
One is AI LED vulnerability 
discovery. 

452
00:24:13,360 --> 00:24:16,160
So Microsoft plans to 
incorporate advanced AI models 

453
00:24:16,600 --> 00:24:18,640
like Claude Mithos. 
Preview through project. 

454
00:24:18,640 --> 00:24:22,880
Glasswing directly into their 
security development life cycle 

455
00:24:23,000 --> 00:24:26,600
to identify vulnerabilities and 
develop medications early in the

456
00:24:26,600 --> 00:24:29,440
process. 
That were previously possible. 

457
00:24:29,560 --> 00:24:33,320
AI assisted discoveries will 
flow through existing MSRC 

458
00:24:33,320 --> 00:24:35,480
processes, including Patch 
Tuesday. 

459
00:24:35,920 --> 00:24:37,880
So the. 
Distribution mechanism stays 

460
00:24:37,880 --> 00:24:41,280
predictable even though the 
discovery model becomes more 

461
00:24:41,280 --> 00:24:43,800
powerful. 
I totally said that earlier in 

462
00:24:43,800 --> 00:24:47,400
the the thing and I did not mean
to, but that that was exactly my

463
00:24:47,400 --> 00:24:51,400
point is, even though there may 
be more discoveries, the way we 

464
00:24:51,400 --> 00:24:53,880
deal with them remains the same 
and it's consistent and 

465
00:24:53,880 --> 00:24:55,640
predictable. 
And it's that second Tuesday of 

466
00:24:55,640 --> 00:24:57,440
the month you're going to get 
your Windows updates. 

467
00:24:57,440 --> 00:25:01,360
And that that should be 
reassuring to everyone that 

468
00:25:01,760 --> 00:25:05,120
unless it rises to that need of 
this is going to be an out of 

469
00:25:05,120 --> 00:25:08,280
band patch, it's going to be 
through that same process. 

470
00:25:08,280 --> 00:25:11,360
So if you're really good, you're
using your Windows auto patch, 

471
00:25:11,360 --> 00:25:14,760
you're using Windows hot patch, 
you're all set to deploy patches

472
00:25:14,760 --> 00:25:16,880
at scale through deployment 
rings and you're not 

473
00:25:16,880 --> 00:25:19,440
individually approving every 
patch, which my God, why are you

474
00:25:19,440 --> 00:25:21,440
still doing that? 
You're in really good shape now.

475
00:25:21,440 --> 00:25:25,440
If you still have some bumps in 
every patch, Tuesday's painful. 

476
00:25:25,920 --> 00:25:29,000
Maybe now is a good time to try 
to streamline that process. 

477
00:25:29,400 --> 00:25:31,880
And this is something that we 
also talked about South in this 

478
00:25:31,880 --> 00:25:34,640
blog. 
Microsoft says that they're 

479
00:25:34,640 --> 00:25:37,160
going to use AI to proactively 
scan open. 

480
00:25:37,160 --> 00:25:40,160
Source code bases which? 
Will identify issues handled 

481
00:25:40,160 --> 00:25:42,000
through coordinated 
vulnerability disclosure. 

482
00:25:42,320 --> 00:25:43,880
You made that point, Andy. 
Yeah. 

483
00:25:43,880 --> 00:25:46,760
And so like, like I said, 
because we have access to the 

484
00:25:46,760 --> 00:25:51,680
stuff and Microsoft is also 
using open source code, you 

485
00:25:51,680 --> 00:25:54,840
know, obviously they're going to
scan the code that you're using 

486
00:25:54,840 --> 00:25:56,640
and then, you know, try to 
coordinate and get those. 

487
00:25:56,640 --> 00:25:59,600
Fixed because they're. 
Using it in the code base so. 

488
00:26:01,280 --> 00:26:04,400
Yep, and I think you'll see the 
Apples and and Googles of the 

489
00:26:04,400 --> 00:26:07,240
world doing the same where any 
libraries they're importing into

490
00:26:07,240 --> 00:26:10,440
their code. 
Obviously Mac OS and iOS are 

491
00:26:10,440 --> 00:26:14,640
both Unix derivative systems. 
They're importing a lot of open 

492
00:26:14,640 --> 00:26:18,480
source modules, and Android's 
built on Linux, so they also 

493
00:26:18,480 --> 00:26:21,720
have a vested interest in making
sure that all those downstream 

494
00:26:21,720 --> 00:26:25,680
components are being scanned, 
not just their closed source 

495
00:26:25,680 --> 00:26:29,320
code as well. 
And then the second pillar is AI

496
00:26:29,320 --> 00:26:33,000
ready posture management. 
So Microsoft has identified 5 

497
00:26:33,120 --> 00:26:37,520
dimensions where AI driven 
attacks gain disproportionate 

498
00:26:37,520 --> 00:26:40,960
advantages. 
Patching open source software 

499
00:26:41,440 --> 00:26:43,280
customer source. 
Code. 

500
00:26:43,920 --> 00:26:46,960
Internet facing assets and 
baseline security hygiene. 

501
00:26:47,040 --> 00:26:51,040
So there's a new Secure Now 
Blade, and this is what you're 

502
00:26:51,040 --> 00:26:54,600
talking about in the Microsoft 
Security Exposure Management, 

503
00:26:55,040 --> 00:26:59,760
which is available today at 
security.microsoft.com/secure 

504
00:26:59,760 --> 00:27:03,040
Now, which gives customers 
guidance and tooling to access 

505
00:27:03,040 --> 00:27:06,680
their current state, prioritize 
actions and models. 

506
00:27:06,680 --> 00:27:09,640
The quote UN quote what if 
scenarios and then applies 

507
00:27:09,640 --> 00:27:11,760
automation to remediate at 
scale. 

508
00:27:12,480 --> 00:27:16,400
The tool set will span attack 
surface discoveries via 

509
00:27:16,560 --> 00:27:20,120
Defender, external attack 
surface management, code 

510
00:27:20,120 --> 00:27:23,840
scanning via Github's Advanced 
Security with Copilot Auto Fix 

511
00:27:23,880 --> 00:27:26,800
and Microsoft Baseline Security 
models for applying foundational

512
00:27:26,800 --> 00:27:30,600
controls across the M3. 65 
services with. 

513
00:27:30,920 --> 00:27:33,680
Impact simulation before 
enforcement. 

514
00:27:34,440 --> 00:27:37,640
So that's pretty cool. 
Yeah, the Secure Now Blade in 

515
00:27:37,640 --> 00:27:40,760
security exposure management is 
not for sale. 

516
00:27:40,880 --> 00:27:44,760
We don't sell that. 
It's included with any level of 

517
00:27:44,760 --> 00:27:46,360
usage for Microsoft security 
products. 

518
00:27:46,360 --> 00:27:49,560
So that's why I was making the 
point this is not a new revenue 

519
00:27:49,560 --> 00:27:51,360
generating announcement or 
anything like that. 

520
00:27:51,360 --> 00:27:53,640
This is baked into existing 
features. 

521
00:27:53,680 --> 00:27:56,080
You cannot buy Microsoft 
security exposure management. 

522
00:27:56,080 --> 00:27:58,400
We don't sell it. 
It's just included with anything

523
00:27:58,400 --> 00:27:59,760
you're using in the Defender 
portal. 

524
00:27:59,920 --> 00:28:02,600
So you can go over there. 
And the more tools you have on 

525
00:28:02,600 --> 00:28:04,000
boarded, the more valuable this 
gets. 

526
00:28:04,000 --> 00:28:05,840
For sure. 
If you're using things like 

527
00:28:05,840 --> 00:28:07,720
external attack service 
management, you get more 

528
00:28:07,720 --> 00:28:10,120
visibility than if you're not. 
But none of those are 

529
00:28:10,120 --> 00:28:14,320
requirements to take advantage 
of this Secure Now Blade and to 

530
00:28:14,320 --> 00:28:18,400
go check it out. 
So security.microsoft.com slash 

531
00:28:18,400 --> 00:28:21,040
secure now. 
And then finally. 

532
00:28:21,440 --> 00:28:24,520
The third pillar is. 
AI powered solutions at 

533
00:28:24,520 --> 00:28:27,320
enterprise scale. 
And so Microsoft plans to 

534
00:28:27,440 --> 00:28:31,880
produce an internal internally 
developed multi model AI driven 

535
00:28:31,880 --> 00:28:35,080
scanning harness and it will 
make it available to customers 

536
00:28:35,080 --> 00:28:40,200
with preview expected in June 
2026, which is like a month 

537
00:28:40,200 --> 00:28:41,040
away. 
Yeah. 

538
00:28:41,760 --> 00:28:46,360
So it internally developed multi
model AI driven scanning harness

539
00:28:46,360 --> 00:28:50,040
and it's going to be productized
and made available to customers.

540
00:28:50,040 --> 00:28:53,480
So that's sometimes that's where
stuff comes from at Microsoft. 

541
00:28:53,480 --> 00:28:55,520
A lot of times is we will 
scratch our own itch. 

542
00:28:55,520 --> 00:28:57,760
Obviously being a company of 
software developers make 

543
00:28:57,760 --> 00:29:00,360
something to help ourselves and 
then say, you know, customers 

544
00:29:00,360 --> 00:29:03,160
would like this too. 
We should we should turn this 

545
00:29:03,160 --> 00:29:04,400
into a product and go sell it to
them. 

546
00:29:04,720 --> 00:29:07,200
Microsoft Sentinel, as an 
example, was originally 

547
00:29:07,200 --> 00:29:10,000
developed in the Xbox 
organization and then was turned

548
00:29:10,000 --> 00:29:14,680
into a full on product that now 
is used by hundreds of thousands

549
00:29:14,680 --> 00:29:17,520
of customers around the world. 
So this is another example. 

550
00:29:17,520 --> 00:29:20,560
And harnesses are always, I 
think the current 

551
00:29:20,560 --> 00:29:23,120
state-of-the-art where the most 
interesting development is 

552
00:29:23,120 --> 00:29:26,680
happening as a lot of these 
models have reached relative 

553
00:29:26,680 --> 00:29:29,600
parity with one another and 
they're all pretty gosh darn 

554
00:29:29,600 --> 00:29:33,200
good and amazing. 
You know, tomato, Tomato GPT 5.5

555
00:29:33,200 --> 00:29:36,360
or Opus 4.7 or whatever. 
The harness is where the 

556
00:29:36,360 --> 00:29:40,400
innovation is because this is 
the part that actually spawns 

557
00:29:40,400 --> 00:29:44,520
like all the, the agents that 
are issuing multiple prompts. 

558
00:29:44,520 --> 00:29:47,800
So you issue a prompt into a 
harness and it's like I need 10 

559
00:29:47,800 --> 00:29:49,640
agents spun up. 
Each of these agents is going to

560
00:29:49,640 --> 00:29:52,840
run these prompts in these 
chains, in these orders to go 

561
00:29:52,840 --> 00:29:55,760
accomplish a given task. 
And so you think you've issued 

562
00:29:55,760 --> 00:29:58,800
one prompt, but under the 
covers, potentially hundreds if 

563
00:29:58,800 --> 00:30:00,440
not thousands of prompts will be
issued. 

564
00:30:00,680 --> 00:30:04,280
That's why these models like 
Copilot Co work, clawed Co work 

565
00:30:04,280 --> 00:30:07,160
open claw. 
They're harnesses and they, they

566
00:30:07,160 --> 00:30:11,880
generate an exponential increase
in, in token consumption 

567
00:30:12,160 --> 00:30:15,200
compared to just issuing like 
traditional prompt and response 

568
00:30:15,200 --> 00:30:18,040
models, like a chachi BT type 
experience. 

569
00:30:18,080 --> 00:30:20,640
And so that's, that's the thing 
to keep in mind is this is 

570
00:30:20,640 --> 00:30:23,120
potentially really, really 
powerful, really, really 

571
00:30:23,120 --> 00:30:27,120
interesting being a, a harness 
that's going to leverage 

572
00:30:27,280 --> 00:30:29,840
multiple models. 
Also to your point, Andy, what 

573
00:30:29,840 --> 00:30:35,080
if this says, hey, Opus 4.7 and 
GPT 5.5, both of you go bang on 

574
00:30:35,080 --> 00:30:37,160
this and tell me what you come 
back with and then compare notes

575
00:30:37,160 --> 00:30:39,560
with each other. 
Incredibly powerful and 

576
00:30:39,560 --> 00:30:42,920
something you can really only 
get from a Microsoft as opposed 

577
00:30:42,920 --> 00:30:45,640
to open AI or, and Tropic's not 
going to offer that 'cause 

578
00:30:45,640 --> 00:30:47,720
they're going to be just focused
on just their models. 

579
00:30:47,720 --> 00:30:50,640
So I think that's also 
interesting when you can go to 

580
00:30:50,640 --> 00:30:53,920
that multi model, multi vendor 
approach within a single tool 

581
00:30:53,920 --> 00:30:56,920
from a single vendor is kind of 
uniquely differentiated. 

582
00:30:56,920 --> 00:30:59,840
So I look forward to hearing 
more about that and gosh, next 

583
00:30:59,840 --> 00:31:03,000
month already exciting times for
security researchers. 

584
00:31:03,880 --> 00:31:05,960
Yeah, in in the article it says 
that. 

585
00:31:05,960 --> 00:31:09,240
The goal is to pair. 
Model output with contacts and 

586
00:31:09,240 --> 00:31:13,080
prioritize prioritization so 
that large volumes of findings 

587
00:31:13,080 --> 00:31:15,400
don't just overwhelm development
teams. 

588
00:31:15,720 --> 00:31:18,720
The solutions designed to 
deliver actionable outcomes and 

589
00:31:18,720 --> 00:31:22,000
not just raw results and then 
rapidly deployed defender 

590
00:31:22,000 --> 00:31:23,640
detections. 
Or. 

591
00:31:23,800 --> 00:31:27,000
AI discovered vulnerabilities 
will also ship alongside the 

592
00:31:27,000 --> 00:31:30,040
corresponding security updates. 
So protection and disclosure 

593
00:31:30,040 --> 00:31:33,160
will happen together. 
And then they did talk about 

594
00:31:33,280 --> 00:31:35,080
because I think this. 
Was released, right? 

595
00:31:35,080 --> 00:31:39,640
Around the same time that Claude
Mythos was announced as well. 

596
00:31:39,640 --> 00:31:43,040
So Microsoft says that through 
Project Glasswing, they're 

597
00:31:43,040 --> 00:31:46,360
working closely with Anthropic 
to test cloud Mythos preview and

598
00:31:46,360 --> 00:31:48,720
then identify, mitigate 
vulnerabilities early, 

599
00:31:48,720 --> 00:31:51,560
coordinate this defense response
across the industry. 

600
00:31:51,600 --> 00:31:56,760
Microsoft is evaluating Mythos 
using a CTI Dash realm, their 

601
00:31:56,760 --> 00:32:00,320
open source benchmark for real 
world detection engineering 

602
00:32:00,320 --> 00:32:03,600
tasks and saw substantial 
improvements relative to prior 

603
00:32:03,600 --> 00:32:06,160
models. 
And then also importantly 

604
00:32:06,160 --> 00:32:09,440
enough, Microsoft is taking 
deliberately a multi model 

605
00:32:09,440 --> 00:32:11,080
approach. 
So they're evaluating models 

606
00:32:11,080 --> 00:32:13,800
from multiple providers and not 
a single model defining their 

607
00:32:13,800 --> 00:32:15,320
strategy. 
And that you know, we talked 

608
00:32:15,320 --> 00:32:19,560
about that here with the chat G.
PTGPT. 5.5 and and I'm sure 

609
00:32:19,560 --> 00:32:23,280
there's some other model that 
maybe Open AI has that that 

610
00:32:23,520 --> 00:32:26,640
maybe hasn't been announced so 
but yeah overall like you know 

611
00:32:27,360 --> 00:32:29,840
throughout this entire. 
Episode it's it's really just. 

612
00:32:29,840 --> 00:32:32,720
About one AI vulnerability 
scanning and how that's speeding

613
00:32:32,720 --> 00:32:35,600
things up, and then how like 
defenders can take advantage. 

614
00:32:35,640 --> 00:32:39,920
Of that right, so. 
I think we're, we're rapidly 

615
00:32:39,960 --> 00:32:42,040
into this. 
I think the technology is 

616
00:32:42,040 --> 00:32:44,320
normalizing. 
We're still in a area of 

617
00:32:44,320 --> 00:32:47,000
innovation. 
But we're really. 

618
00:32:47,160 --> 00:32:50,400
Discovering how we can use. 
AI for security. 

619
00:32:50,400 --> 00:32:55,160
In a meaningful way. 
And yeah, if if you haven't 

620
00:32:55,160 --> 00:32:58,960
really like jumped on the AI 
train, I think now is a good 

621
00:32:58,960 --> 00:33:02,120
time to really start diving in 
and we're going to have a guest 

622
00:33:02,120 --> 00:33:03,920
on in a couple. 
Of weeks to really talk about. 

623
00:33:04,800 --> 00:33:09,160
Some more creative ways to use 
AI for you know, if if you're a 

624
00:33:09,160 --> 00:33:13,080
security person. 
But yeah, this is this. 

625
00:33:13,080 --> 00:33:16,360
Is all to me extremely 
interesting because it's you can

626
00:33:16,360 --> 00:33:19,840
literally see the industry kind 
of evolving in front of your 

627
00:33:19,840 --> 00:33:21,280
eyes, right? 
Yeah. 

628
00:33:21,280 --> 00:33:25,600
I, I mean, the closest in analog
to for most of us would be the 

629
00:33:25,600 --> 00:33:29,920
smartphone revolution in the 
late thousands and early 20s 

630
00:33:29,920 --> 00:33:33,720
when there was an incredible 
pace of innovation there from 

631
00:33:33,720 --> 00:33:37,760
2007 to, I don't know, maybe 
let's say 2015, Maybe for that 

632
00:33:37,760 --> 00:33:42,400
eight-year stretch or so, when 
every year the new Android 

633
00:33:42,400 --> 00:33:45,760
phones and the new iPhones were 
exponential leaps year over 

634
00:33:45,760 --> 00:33:47,480
year. 
They got so much faster, the 

635
00:33:47,480 --> 00:33:52,080
cameras got so much better and 
then they you know, it levelled 

636
00:33:52,080 --> 00:33:54,000
off right. 
Today we say, well, it's boring.

637
00:33:54,000 --> 00:33:57,680
The new iphone's hardly any 
different and you go, you go 

638
00:33:57,680 --> 00:34:00,760
look at like a picture from 
taking on a smartphone in O 

639
00:34:00,760 --> 00:34:04,160
eight O 9 and it's clear, 
immediate smartphone picture, 

640
00:34:04,280 --> 00:34:07,960
you know, and then we reached a 
point and it was about the mid 

641
00:34:07,960 --> 00:34:10,000
twenty 10s when all of a sudden 
they all look pretty good. 

642
00:34:10,560 --> 00:34:13,040
And you you can't dramatically 
go well, that's clearly from an 

643
00:34:13,040 --> 00:34:15,320
iPhone 17 versus that's an 
iPhone 15. 

644
00:34:15,320 --> 00:34:17,920
Like you. 
You can't tell easily unless 

645
00:34:17,920 --> 00:34:19,360
you're a absolute guru in this 
stuff. 

646
00:34:19,360 --> 00:34:21,480
Because the the rate of change 
leveled off, right? 

647
00:34:21,480 --> 00:34:25,040
The the rate of innovation 
slowed as a lot of the easier 

648
00:34:25,040 --> 00:34:26,960
wins all got rung out of the 
system. 

649
00:34:26,960 --> 00:34:31,080
And we're still in that phase of
rapid innovation and rapid 

650
00:34:31,080 --> 00:34:34,000
change right now. 
But it's we are starting to see 

651
00:34:34,280 --> 00:34:36,880
the levelling off, right. 
We're starting to see who are 

652
00:34:36,880 --> 00:34:41,400
clearly the players. 
Is there going to be a lasting 

653
00:34:41,400 --> 00:34:44,199
differentiated Moat in terms of 
frontier model capability? 

654
00:34:44,400 --> 00:34:47,080
Probably not. 
It seems to be about a three, 

655
00:34:47,080 --> 00:34:48,239
four horse race. 
Maybe. 

656
00:34:48,760 --> 00:34:50,920
You've got you've got your 
googles, you've got your open 

657
00:34:50,920 --> 00:34:53,400
AIS, you've got your anthropics,
Microsoft AI. 

658
00:34:53,400 --> 00:34:56,040
Do not sleep on us. 
We have an incredible team with 

659
00:34:56,040 --> 00:35:00,320
Mustafa Soleiman leading the way
and we can literally have the 

660
00:35:00,320 --> 00:35:04,920
rights to leverage open AIS IP 
to train our own models. 

661
00:35:04,920 --> 00:35:07,680
That was in the announcement 
earlier in the week actually. 

662
00:35:07,720 --> 00:35:09,680
And if you don't haven't 
followed frontier model 

663
00:35:09,680 --> 00:35:11,880
development, it's incredibly 
iterative. 

664
00:35:12,240 --> 00:35:15,440
It's you, you distill your 
previous model to make your next

665
00:35:15,440 --> 00:35:18,200
model in some ways. 
And you you use your old model 

666
00:35:18,200 --> 00:35:21,200
to train your new model. 
So like the first model was the 

667
00:35:21,200 --> 00:35:23,640
hardest one to make. 
And then they get easier in in 

668
00:35:23,640 --> 00:35:26,360
scare quotes, right? 
But with time as you're able to 

669
00:35:26,360 --> 00:35:28,680
kind of draft off your previous 
learnings. 

670
00:35:28,680 --> 00:35:33,360
Well, effectively Microsoft gets
to draft off of open AI as all 

671
00:35:33,360 --> 00:35:35,480
their work on to create a 
frontier model. 

672
00:35:35,480 --> 00:35:40,240
And, and we very openly said Mai
is coming with a frontier model 

673
00:35:40,240 --> 00:35:43,960
of its own and it's already 
shipped speech generation model.

674
00:35:43,960 --> 00:35:46,720
That's incredible image 
generation models that are 

675
00:35:46,720 --> 00:35:50,040
really, really darn good. 
And so there's more to come in 

676
00:35:50,040 --> 00:35:51,600
this space too. 
So I would not sleep in 

677
00:35:51,600 --> 00:35:54,440
Microsoft either. 
But anyhow, the point is, I 

678
00:35:54,440 --> 00:35:57,480
think you summarized it well, 
Andy, that yes, AI is 

679
00:35:57,480 --> 00:36:01,480
compressing this timeline from 
discovery to exploitation to 

680
00:36:01,480 --> 00:36:03,160
patching. 
And there will be a period of 

681
00:36:03,160 --> 00:36:06,760
time where we have kind of I 
think a more intensive patching 

682
00:36:07,360 --> 00:36:09,800
process for a while. 
Then I think we emerge the other

683
00:36:09,800 --> 00:36:13,280
side and that levels off. 
However, now technologies like 

684
00:36:13,280 --> 00:36:17,280
Claude Security are bringing 
this availability not just to a 

685
00:36:17,280 --> 00:36:20,160
select few, but to all 
enterprises that are willing to,

686
00:36:20,280 --> 00:36:23,280
you know, buy in. 
And then our employer, Microsoft

687
00:36:23,280 --> 00:36:26,440
is, is weaving AI throughout the
system, not just from 

688
00:36:26,520 --> 00:36:31,600
discovering vulnerabilities, but
AI ready posture management and 

689
00:36:31,600 --> 00:36:34,640
then using AI to enable 
solutions at enterprise scale 

690
00:36:34,640 --> 00:36:39,320
and, and global scale and beyond
in terms of accelerating not 

691
00:36:39,320 --> 00:36:43,880
just the the patching, but the 
detections as well and building 

692
00:36:43,880 --> 00:36:47,560
those into the Defender product 
at the same time as the patch is

693
00:36:47,560 --> 00:36:50,200
built in the windows. 
And that's the benefit you get 

694
00:36:50,200 --> 00:36:52,320
when you develop Defender, 
friend point and Windows side by

695
00:36:52,320 --> 00:36:54,080
side. 
You can do both. 

696
00:36:54,080 --> 00:36:56,840
Let's develop the patch and 
let's develop the detections 

697
00:36:56,840 --> 00:36:59,360
together. 
And so I think you're going to 

698
00:36:59,360 --> 00:37:02,200
see more like that to come from 
everyone. 

699
00:37:02,480 --> 00:37:05,560
But ultimately, I think Andy, 
you and I are both, if you're in

700
00:37:05,560 --> 00:37:08,040
cybersecurity, you have a 
natural skepticism, a healthy 

701
00:37:08,040 --> 00:37:10,160
skepticism. 
And, and it's important that 

702
00:37:10,160 --> 00:37:12,840
everyone keeps that moving 
forward. 

703
00:37:13,360 --> 00:37:17,320
But I really do take a bullish 
positive view on, on where our 

704
00:37:17,320 --> 00:37:20,200
industry is headed. 
And it's been, it's been kind of

705
00:37:20,200 --> 00:37:22,080
dark, man. 
We've had some dark 

706
00:37:22,080 --> 00:37:24,640
conversations about like the 
rate of attacks keep growing. 

707
00:37:24,640 --> 00:37:27,480
And you know, the defensive 
teams are not getting more 

708
00:37:27,480 --> 00:37:29,280
funding, they're not getting 
more budget. 

709
00:37:29,280 --> 00:37:32,080
It's it's getting harder to 
protect all the different 

710
00:37:32,080 --> 00:37:33,280
workloads we're getting to 
protect. 

711
00:37:33,280 --> 00:37:35,160
We need a hero to come save the 
day. 

712
00:37:35,160 --> 00:37:39,080
We need some, some inflection 
point to come, even the scales. 

713
00:37:39,080 --> 00:37:40,680
And all of a sudden that's 
arrived. 

714
00:37:41,120 --> 00:37:44,200
And in some ways, there's a 
certain class of people who want

715
00:37:44,200 --> 00:37:45,640
to look at their nose down on 
it. 

716
00:37:45,920 --> 00:37:50,480
This is not not the silver 
bullet, not the savior, but this

717
00:37:50,480 --> 00:37:54,960
is a tool that can help us kind 
of even even the fight. 

718
00:37:55,400 --> 00:37:58,080
And that is something I'm really
excited about because being in 

719
00:37:58,080 --> 00:38:02,680
cybersecurity, I really kind of 
got in this game in 2017 and 

720
00:38:02,680 --> 00:38:05,400
through about 2022, the story 
just kept getting worse every 

721
00:38:05,400 --> 00:38:07,760
year. 
And now it feels like there is 

722
00:38:07,760 --> 00:38:11,920
some hope that we can turn the 
tides and make this a manageable

723
00:38:11,920 --> 00:38:14,360
problem without burning out 
humanity on it. 

724
00:38:15,280 --> 00:38:17,600
I'm, I'm excited about that. 
Why would I not be? 

725
00:38:18,160 --> 00:38:21,040
So I, I, I choose to take the 
positive viewpoint. 

726
00:38:21,280 --> 00:38:24,240
I am not ignorant to the fact or
have my head in the sand that 

727
00:38:24,240 --> 00:38:25,720
there's going to be pain along 
the way. 

728
00:38:25,720 --> 00:38:28,560
There always is. 
And look at, you know, I, I 

729
00:38:28,560 --> 00:38:30,800
compared this to smartphones. 
Let me tie back to that for a 

730
00:38:30,800 --> 00:38:32,880
second. 
We are dealing with the negative

731
00:38:33,040 --> 00:38:37,840
outcomes from smartphones, from 
social media certainly has I, I 

732
00:38:37,840 --> 00:38:42,560
think impacted our ability to 
have positive discourse and to 

733
00:38:42,560 --> 00:38:45,640
discuss things in a healthy and 
civil way. 

734
00:38:45,640 --> 00:38:49,600
It is impacted that and you have
smartphone addiction that is 

735
00:38:49,600 --> 00:38:52,960
impacting a lot of people, 
shockingly a lot of by the way, 

736
00:38:52,960 --> 00:38:57,000
in my life, in my experience, 
the previous generations, a lot 

737
00:38:57,000 --> 00:39:00,120
of them boomers. 
I see, I see. 

738
00:39:00,160 --> 00:39:02,120
I have several boomers in my 
life who have struggling with 

739
00:39:02,120 --> 00:39:04,760
smartphone addiction today and 
it's a real problem and they 

740
00:39:05,160 --> 00:39:07,640
just dooms going on their phone 
all day long. 

741
00:39:08,080 --> 00:39:11,400
And, and we have the impacts in 
schools with cyberbullying and 

742
00:39:11,400 --> 00:39:15,960
everything else, just with every
positive technology where it is 

743
00:39:15,960 --> 00:39:19,560
amazing that I can beam pictures
of my children to their 

744
00:39:19,560 --> 00:39:22,200
grandparents, you know, with the
snap of a finger and they can 

745
00:39:22,280 --> 00:39:25,000
feel like they're more connected
and watch them grow up in ways 

746
00:39:25,400 --> 00:39:27,280
my grandparents never got to see
with me. 

747
00:39:27,480 --> 00:39:30,400
That's amazing and beautiful and
wonderful and bringing humanity 

748
00:39:30,400 --> 00:39:32,240
together. 
There's been all these negative 

749
00:39:32,240 --> 00:39:34,560
impacts too. 
This will be no different. 

750
00:39:34,560 --> 00:39:37,600
There will be downsides to this,
but I think there are tremendous

751
00:39:37,600 --> 00:39:39,640
upsides too. 
And I'm choosing to focus on 

752
00:39:39,640 --> 00:39:41,680
those. 
And I think this show is a great

753
00:39:41,680 --> 00:39:44,320
example of all the positive 
things that AI can bring to 

754
00:39:44,320 --> 00:39:46,360
security defenders lives now and
in the future. 

755
00:39:46,720 --> 00:39:49,680
Great wrap up there. 
Adam, all right, well, thanks 

756
00:39:49,680 --> 00:39:51,400
for watching and listening as 
always. 

757
00:39:51,400 --> 00:39:52,760
That's. 
Our show for this week. 

758
00:39:53,320 --> 00:39:55,920
Our contact information as well 
as the links. 

759
00:39:55,920 --> 00:39:58,840
To the articles that we used for
the show will. 

760
00:39:58,840 --> 00:40:01,360
Be in the show notes. 
If you have any questions or 

761
00:40:01,360 --> 00:40:04,440
topics you want us to talk about
in the future, just e-mail us. 

762
00:40:04,920 --> 00:40:06,120
Thanks. 
We'll talk to you guys next 

763
00:40:06,120 --> 00:40:08,120
week. 
Thank you for listening to the 

764
00:40:08,120 --> 00:40:10,960
Blue Security Podcast. 
Please check out the show notes,

765
00:40:10,960 --> 00:40:13,760
catch up on episodes you may 
have missed, and subscribe so 

766
00:40:13,760 --> 00:40:15,520
you don't miss any future 
episodes. 

767
00:40:15,560 --> 00:40:20,200
Find Andy on Twitter at a Jaw 
Zero and Adam at AJ Brewer. 

768
00:40:20,440 --> 00:40:22,040
See you at our next episode.
