1
00:00:00,040 --> 00:00:04,440
So last week Anthropic, so the 
maker of Claude, fell out with 

2
00:00:04,440 --> 00:00:07,080
the US government and the 
Department of War. 

3
00:00:07,200 --> 00:00:09,600
This is a huge high profile 
fallout. 

4
00:00:09,840 --> 00:00:14,000
Trump even took to true social 
to claim that Anthropic and 

5
00:00:14,000 --> 00:00:19,840
Claude are enemies of peace and 
the government and huge woke 

6
00:00:19,840 --> 00:00:24,440
liberals and all because Claude 
said no to some really quite 

7
00:00:24,440 --> 00:00:27,200
terrifying use cases that the 
army wanted to implement. 

8
00:00:27,440 --> 00:00:30,720
And then the very same 
afternoon, Open AI announced a 

9
00:00:30,720 --> 00:00:33,680
high profile partnership with 
the Department of War. 

10
00:00:33,840 --> 00:00:40,400
And since then, we've seen Open 
AI uninstalls rise by 295%, one 

11
00:00:40,400 --> 00:00:44,480
star reviews spike and double 
day over day. 

12
00:00:44,640 --> 00:00:47,760
And then, despite being banned 
by the government, Anthropic's 

13
00:00:47,760 --> 00:00:51,640
cord was then used to conduct 
air strikes against Iran. 

14
00:00:51,760 --> 00:00:54,960
What a crazy turn of events. 
And today I'm covering it all. 

15
00:00:55,120 --> 00:00:58,440
What the actual contracts say 
and the fine print now, the 

16
00:00:58,440 --> 00:01:01,960
things that most people don't 
know about the situation, the 

17
00:01:01,960 --> 00:01:06,680
specific details that really do 
matter and trying to understand 

18
00:01:06,680 --> 00:01:10,200
the answer to a key question. 
When technology becomes so 

19
00:01:10,200 --> 00:01:13,760
powerful, who has the right to 
decide how to use it? 

20
00:01:14,640 --> 00:01:15,960
This is in a Leaf with Jack 
Horton. 

21
00:01:16,440 --> 00:01:30,560
Hope you enjoy the show. 
So on Thursday evening, on 

22
00:01:30,560 --> 00:01:36,560
February 27th, Sam Altman sent 
an internal memo to every single

23
00:01:36,560 --> 00:01:39,840
employee Open AI. 
So the Wall Street Journal broke

24
00:01:39,840 --> 00:01:42,160
the story. 
And in it, Altman actually said 

25
00:01:42,160 --> 00:01:46,320
the open AI had long believed 
that AI should not ever be used 

26
00:01:46,320 --> 00:01:49,880
for mass surveillance or 
autonomous lethal weapons. 

27
00:01:50,160 --> 00:01:53,600
And this is a case where it's 
important to me. 

28
00:01:53,600 --> 00:01:55,560
And I quote, this is by Sam 
Altman. 

29
00:01:55,880 --> 00:01:58,960
This is a case where it's 
important to me that we do the 

30
00:01:58,960 --> 00:02:02,080
right thing and not the easy 
thing that looks disingenuous. 

31
00:02:02,200 --> 00:02:06,680
And less than 24 hours later, 
Open AI then signed a deal with 

32
00:02:06,680 --> 00:02:10,320
the Pentagon and therefore the 
Department of War to deploy its 

33
00:02:10,320 --> 00:02:13,320
models in throughout its 
security operations. 

34
00:02:13,440 --> 00:02:16,000
So to understand how we got 
here, I think you need to 

35
00:02:16,000 --> 00:02:19,720
understand what happened in the 
days before this agreement, 

36
00:02:19,920 --> 00:02:22,680
because Anthropic had been 
embedded in the Pentagon system 

37
00:02:22,680 --> 00:02:27,200
since, you know, June 2024, and 
it was the first frontier AI 

38
00:02:27,200 --> 00:02:30,360
model to get there. 
They got into this partnership 

39
00:02:30,360 --> 00:02:35,400
via Palantir and AWS and 
actually got security level 

40
00:02:35,400 --> 00:02:39,880
clearance at level 6, which is 
classified as secret clearance. 

41
00:02:40,280 --> 00:02:43,880
And by the summer of 2025, they 
had a contract worth up to $200 

42
00:02:43,880 --> 00:02:47,000
million. 
And Claude was being used by 

43
00:02:47,000 --> 00:02:50,360
CENTCOM for intelligence 
assessments, identifying 

44
00:02:50,360 --> 00:02:53,320
targets, simulating battle 
scenarios. 

45
00:02:53,440 --> 00:02:55,360
And it was reportedly used 
during the capture of the 

46
00:02:55,360 --> 00:02:58,480
Venezuelan president in January 
2026. 

47
00:02:58,920 --> 00:03:00,960
And, you know, this wasn't a 
company sitting on the sidelines

48
00:03:00,960 --> 00:03:04,400
here being super ethical. 
Anthropic was already very deep 

49
00:03:04,400 --> 00:03:07,560
inside the US military 
establishment, far more than any

50
00:03:07,560 --> 00:03:10,240
other AI lab. 
Now the dispute that really 

51
00:03:10,240 --> 00:03:13,360
kicked off publicly and had 
Trump weighing in, calling 

52
00:03:13,360 --> 00:03:16,440
Anthropic an enemy of the 
people, actually was triggered 

53
00:03:16,440 --> 00:03:20,000
by Palentia. 
Palentia tried to get Claude or 

54
00:03:20,000 --> 00:03:23,480
Anthropic to agree to what they 
call an all lawful use standard,

55
00:03:23,760 --> 00:03:26,000
meaning that essentially the 
military could use claw for 

56
00:03:26,000 --> 00:03:29,120
anything they wanted that wasn't
explicitly illegal. 

57
00:03:30,040 --> 00:03:32,600
And Anthropic said, yeah, that's
absolutely fine, apart from 2 

58
00:03:32,600 --> 00:03:36,560
exceptions that it wouldn't 
budget on, which was #1 mass 

59
00:03:36,560 --> 00:03:40,440
surveillance or mass domestic 
surveillance of the American 

60
00:03:40,440 --> 00:03:43,920
people and #2 was fully 
autonomous weaponry that would 

61
00:03:43,920 --> 00:03:45,920
take humans completely out of 
the loop. 

62
00:03:46,760 --> 00:03:49,440
Everything else, as I said, was 
completely on the table. 

63
00:03:49,520 --> 00:03:51,680
All fine. 
Every other military use case 

64
00:03:51,680 --> 00:03:53,560
you can think of, just not those
two. 

65
00:03:53,760 --> 00:03:58,560
Now on Tuesday the 25th of 
February, the Defence Secretary 

66
00:03:58,760 --> 00:04:02,520
of the US government met with 
the CEO of Anthropic, Dario 

67
00:04:02,520 --> 00:04:05,160
Amadei at the Pentagon and 
essentially they gave each of 

68
00:04:05,160 --> 00:04:08,080
their deadline which was Friday.
By 5:00 PM we have to have 

69
00:04:08,080 --> 00:04:11,400
agreed to all lawful use or lose
the contract completely. 

70
00:04:11,800 --> 00:04:14,080
So lose that massive contract 
with the US government. 

71
00:04:14,680 --> 00:04:17,160
And that night the Pentagon sent
what it called it's last and 

72
00:04:17,160 --> 00:04:19,560
final offer. 
So Anthropics team had reviewed 

73
00:04:19,560 --> 00:04:22,680
the offer that was given to them
and essentially their conclusion

74
00:04:22,680 --> 00:04:25,680
was, and all the lawyers 
conclusion was that there was 

75
00:04:25,680 --> 00:04:29,760
lots of wording that frames the 
agreement as a compromise, but 

76
00:04:29,760 --> 00:04:34,560
ultimately left all the legal 
loopholes to just do whatever 

77
00:04:34,560 --> 00:04:36,360
they want anyway. 
So essentially we had to just 

78
00:04:36,360 --> 00:04:40,160
disregard all safeguards. 
And it got that heated that the 

79
00:04:40,160 --> 00:04:44,040
CEO of Anthropic, Amede, 
published his response on the 

80
00:04:44,040 --> 00:04:45,640
Anthropic website. 
You can go see that today. 

81
00:04:45,920 --> 00:04:48,720
His arguments was specific and 
not ideological. 

82
00:04:49,040 --> 00:04:52,840
He said on weapons, current 
frontier AI models are just not 

83
00:04:52,840 --> 00:04:55,840
reliably good enough to engage 
in targets, and deploying them 

84
00:04:55,840 --> 00:04:58,680
are just going to endanger 
American fighters and civilians.

85
00:04:59,120 --> 00:05:01,680
And on surveillance, he said 
that there is no federal law 

86
00:05:01,920 --> 00:05:05,640
that stops AI conducting mass 
surveillance on millions of 

87
00:05:05,640 --> 00:05:08,480
people. 
But federal agencies haven't yet

88
00:05:08,480 --> 00:05:12,280
caught up with laws to protect 
people from mass surveillance. 

89
00:05:12,440 --> 00:05:16,600
You know, data brokers can 
legally sell mass data to the US

90
00:05:16,600 --> 00:05:19,800
government and then use those 
models to quite literally 

91
00:05:20,400 --> 00:05:23,280
conduct huge mass surveillance. 
And if you think about AI 

92
00:05:23,280 --> 00:05:26,640
models, what they're fantastic 
about is turning huge amounts of

93
00:05:26,640 --> 00:05:32,120
data into the ability to zone in
on particular people, groups of 

94
00:05:32,120 --> 00:05:35,080
people or situations. 
You know, before AI, think about

95
00:05:35,080 --> 00:05:38,840
this, an agency that buys and 
sells people's data, it was 

96
00:05:38,840 --> 00:05:41,680
basically like pouring data into
a big bucket. 

97
00:05:41,960 --> 00:05:43,920
And it's really difficult to 
make sense of that data. 

98
00:05:43,920 --> 00:05:46,400
Yes, you can use it, and it's 
really powerful for machine 

99
00:05:46,400 --> 00:05:49,440
learning models, and there's 
loads of nefarious, dangerous 

100
00:05:49,440 --> 00:05:51,640
use cases there. 
But with AI, that same data 

101
00:05:51,640 --> 00:05:54,200
becomes an insane surveillance 
system. 

102
00:05:54,560 --> 00:05:57,000
You know, you can cross 
reference, you can automatically

103
00:05:57,000 --> 00:06:00,440
build profiles that flags 
individuals or tracks any 

104
00:06:00,440 --> 00:06:03,840
pattern you want across millions
and millions of people 

105
00:06:04,000 --> 00:06:06,400
instantaneously. 
The technology is what makes 

106
00:06:06,400 --> 00:06:08,040
that data really, really 
dangerous. 

107
00:06:08,400 --> 00:06:10,640
And that's the gap that 
Anthropic took a stance on. 

108
00:06:10,800 --> 00:06:14,320
And I have to say, I've been a 
big cloud user for a long time. 

109
00:06:14,400 --> 00:06:16,080
I couldn't even imagine using 
open AI. 

110
00:06:16,080 --> 00:06:18,520
I've not got an open AI 
subscription. 

111
00:06:18,520 --> 00:06:21,400
I tell everyone to not use open 
AI for a number of reasons. 

112
00:06:21,400 --> 00:06:25,560
We'll discuss those, I'm sure. 
But it makes me respect the 

113
00:06:25,560 --> 00:06:28,840
company a bit more as well 
because Amadei wrote and I 

114
00:06:28,840 --> 00:06:31,200
quote, these threats do not 
change our position. 

115
00:06:31,520 --> 00:06:35,080
We cannot in good conscience 
accede their request. 

116
00:06:35,200 --> 00:06:39,320
You know, anyone that's been a 
entrepreneur that has their own 

117
00:06:39,320 --> 00:06:43,240
business or is running high 
profile, high pressure teams, 

118
00:06:43,240 --> 00:06:47,360
for example, knows that the 
pressure to say yes to contract 

119
00:06:47,360 --> 00:06:49,120
can be pretty overwhelming at 
times. 

120
00:06:49,120 --> 00:06:52,120
Because although a company from 
the outside looks amazing, 

121
00:06:52,120 --> 00:06:55,560
they're always, as we all know, 
internal stresses and pressures.

122
00:06:55,920 --> 00:06:59,080
And sometimes doing that right 
thing is not easy, especially 

123
00:06:59,080 --> 00:07:01,720
when you're throwing away 
hundreds and hundreds and 

124
00:07:01,720 --> 00:07:03,840
hundreds of millions and 
probably billions over a couple 

125
00:07:03,840 --> 00:07:08,800
of years of potential revenue. 
And he said no, which I'm so 

126
00:07:08,920 --> 00:07:10,720
relieved up. 
So yeah, that's that's all the 

127
00:07:10,720 --> 00:07:12,880
situation that led up to Donald 
Trump. 

128
00:07:12,880 --> 00:07:16,440
And I quote, calling them left 
wing nut jobs, putting American 

129
00:07:16,440 --> 00:07:19,840
lives at risk and threatens 
actually the full power of the 

130
00:07:19,840 --> 00:07:23,760
presidency with major civil and 
criminal criminal consequences 

131
00:07:24,320 --> 00:07:26,600
if they didn't agree to 
cooperate. 

132
00:07:26,800 --> 00:07:28,840
And you know, a few minutes 
later, they called Anthropic a 

133
00:07:28,840 --> 00:07:31,520
complete supply chamber that 
must be removed from every 

134
00:07:31,520 --> 00:07:50,060
system. 
That same evening, about 8-9 

135
00:07:50,060 --> 00:07:53,520
hours before the US launched 
strikes on Iran, Sam Altman 

136
00:07:53,520 --> 00:07:56,160
announced on X that Open AI had 
reached an agreement with the 

137
00:07:56,160 --> 00:07:57,680
Pentagon. 
Surprise, surprise. 

138
00:07:57,840 --> 00:08:00,320
So yeah, let's, let's get into 
what Open AI's agreement looks 

139
00:08:00,320 --> 00:08:03,920
like, because as I've said, that
they're seeing huge backlash 

140
00:08:03,920 --> 00:08:05,920
right now from their customers 
and user groups. 

141
00:08:06,160 --> 00:08:09,760
On the face of it, both 
companies are claiming identical

142
00:08:09,760 --> 00:08:11,960
principles. 
The difference is the mechanism,

143
00:08:12,120 --> 00:08:14,040
and the mechanism is quite 
literally everything. 

144
00:08:14,360 --> 00:08:19,040
So Anthropic wanted specific 
contractual prohibitions, so 

145
00:08:19,040 --> 00:08:22,720
language in the contract itself 
that specifically banned 

146
00:08:22,720 --> 00:08:26,440
surveillance and autonomous 
weaponry regardless of their 

147
00:08:26,440 --> 00:08:29,920
legal status. 
So whether or not that is made 

148
00:08:29,920 --> 00:08:33,360
legal in the next two, 5-7 
years, they did not want AI to 

149
00:08:33,360 --> 00:08:34,880
be used for that. 
And obviously that's what the 

150
00:08:34,880 --> 00:08:37,520
Pentagon refused to accept. 
Their position was that no 

151
00:08:37,520 --> 00:08:41,240
private company should ever have
veto power over how the military

152
00:08:41,240 --> 00:08:43,520
uses technology. 
You know, the Constitution and 

153
00:08:43,520 --> 00:08:46,200
existing laws are the 
appropriate constraints in the 

154
00:08:46,200 --> 00:08:48,520
view of the Pentagon. 
And, you know, there's an 

155
00:08:48,520 --> 00:08:51,680
argument to say that it's, you 
know, and this is what was on 

156
00:08:51,800 --> 00:08:55,400
all American news channels on 
CBS, saying on some levels, you 

157
00:08:55,400 --> 00:08:57,520
have to trust the military to do
the right thing. 

158
00:08:57,520 --> 00:09:01,520
And the geopolitical argument 
behind, you know, that stance is

159
00:09:01,520 --> 00:09:03,400
real. 
But there are real threats all 

160
00:09:03,400 --> 00:09:05,480
over the world. 
And China's AI labs are 

161
00:09:05,480 --> 00:09:07,880
definitely not negotiating red 
lines with their government. 

162
00:09:08,000 --> 00:09:11,600
Their integration is moving 
incredibly fast because there's 

163
00:09:11,600 --> 00:09:14,240
0 friction. 
And so the Pentagon will view 

164
00:09:14,240 --> 00:09:19,000
this as an arms race with China.
And you can't let from their 

165
00:09:19,000 --> 00:09:21,280
perspective, the government's 
perspective and the Pentagon's 

166
00:09:21,280 --> 00:09:24,840
perspective, the vendors set the
rules, whereas Open AI have 

167
00:09:24,840 --> 00:09:28,440
taken a very different approach.
Instead of any contractual 

168
00:09:28,440 --> 00:09:31,600
prohibitions, they cited 
existing laws and policies. 

169
00:09:31,600 --> 00:09:34,120
So the 4th Amendment, the 
Foreign International 

170
00:09:34,120 --> 00:09:38,840
Surveillance Act, which is a 
Pentagon directive against 

171
00:09:38,920 --> 00:09:41,920
autonomous weaponry, their 
position is essentially, you 

172
00:09:41,920 --> 00:09:44,720
know, these laws already cover 
all of our concerns. 

173
00:09:44,720 --> 00:09:46,880
The government has agreed to 
follow them, and we build 

174
00:09:46,880 --> 00:09:50,400
technical safeguards into the 
model as an additional layer of 

175
00:09:50,400 --> 00:09:51,960
protection. 
But this is where, if you read 

176
00:09:51,960 --> 00:09:55,040
the fine print where, you know, 
there's a gap that opens up. 

177
00:09:55,240 --> 00:09:59,680
So Jessica Tillotman, which is 
she's the associate Dean for 

178
00:09:59,680 --> 00:10:02,280
government procurement law and 
George Washington University, 

179
00:10:02,600 --> 00:10:05,640
she analysed the published 
contract and she concluded that 

180
00:10:05,760 --> 00:10:09,800
it doesn't give open AI an 
anthropic style ability to stop 

181
00:10:09,800 --> 00:10:13,640
the government using the 
technology on those use cases. 

182
00:10:13,640 --> 00:10:16,920
Though it does state that the 
Pentagon cannot use the 

183
00:10:16,920 --> 00:10:20,040
technology when breaking laws. 
Which sounds reassuring until 

184
00:10:20,040 --> 00:10:23,440
you remember that the entire 
reason Anthropic drew its lines 

185
00:10:23,680 --> 00:10:27,640
was that no existing laws 
prohibit the things that we 

186
00:10:27,640 --> 00:10:30,720
should be worried about because 
law has not caught up yet with 

187
00:10:30,720 --> 00:10:33,240
technology. 
You know, the 2023 Pentagon 

188
00:10:33,240 --> 00:10:36,600
directive that was against 
autonomous weaponry, The open AI

189
00:10:36,600 --> 00:10:38,720
citing, you know, it doesn't 
actually ban them. 

190
00:10:38,720 --> 00:10:42,280
All it is is creating guidelines
on how it should and should not 

191
00:10:42,280 --> 00:10:45,480
be used. 
And there's always a waiver that

192
00:10:45,480 --> 00:10:49,720
essentially is a clause stating 
that urgent military needs means

193
00:10:49,720 --> 00:10:50,840
that you can do anything you 
want. 

194
00:10:50,920 --> 00:10:53,720
And Altman even admitted in an 
AMA. 

195
00:10:54,000 --> 00:10:56,520
So it asked me anything on XA 
couple days later. 

196
00:10:56,520 --> 00:11:00,200
The deal was very, very rushed 
and the optics definitely don't 

197
00:11:00,200 --> 00:11:03,280
look good for open AR. 
And he even went further and 

198
00:11:03,280 --> 00:11:06,360
said on Monday that we shouldn't
have rushed to get this out on 

199
00:11:06,360 --> 00:11:08,760
Friday. 
Even open a eyes own team and 

200
00:11:08,760 --> 00:11:12,320
people noticed. 
Leo Gow, an alignment researcher

201
00:11:12,320 --> 00:11:15,360
of the company even said that. 
You know, it's so obvious that 

202
00:11:15,360 --> 00:11:18,120
this is weak and the government 
can now just technically use it 

203
00:11:18,120 --> 00:11:21,080
for anything that is deemed 
lawful, but there are now 

204
00:11:21,080 --> 00:11:24,120
missing laws to deem what is or 
is not lawful when it comes to 

205
00:11:24,120 --> 00:11:47,420
things like mass surveillance. 
Now that there is always a 

206
00:11:47,420 --> 00:11:51,220
counterpart, you know, which is 
opening eye eyes argued that if 

207
00:11:51,220 --> 00:11:53,340
you don't trust the government 
to follow the law, why would you

208
00:11:53,340 --> 00:11:56,860
trust it to honour contract 
terms either, You know, if the 

209
00:11:56,860 --> 00:11:59,580
Pentagon is going to do mass 
surveillance regardless, you 

210
00:11:59,580 --> 00:12:02,140
know, align and drop its 
contract wouldn't stop them 

211
00:12:02,140 --> 00:12:03,460
doing it. 
And there definitely is 

212
00:12:03,460 --> 00:12:06,140
something to that. 
However, contract term shape 

213
00:12:06,140 --> 00:12:09,920
behaviour and it creates 
oversight mechanisms and 

214
00:12:09,920 --> 00:12:12,320
establishes political 
consequences for actions and 

215
00:12:12,320 --> 00:12:14,400
violations. 
You know, a line in the sand 

216
00:12:14,400 --> 00:12:17,320
that someone might just step 
over anyway is better than no 

217
00:12:17,320 --> 00:12:19,440
line in the sand. 
And anyway, I mean, there's been

218
00:12:19,440 --> 00:12:21,440
huge fallout as you can just 
imagine. 

219
00:12:21,720 --> 00:12:24,560
The council ChatGPT movement 
has, it's been going on since 

220
00:12:24,560 --> 00:12:26,640
January. 
I mean, I actually cancelled 

221
00:12:26,640 --> 00:12:31,600
mine in January because I found 
Greg Brockman was the biggest 

222
00:12:31,600 --> 00:12:37,760
donor to Trump, donating over 
$25 million to his action 

223
00:12:37,760 --> 00:12:40,000
committee. 
And I think the Pentagon 

224
00:12:40,400 --> 00:12:43,280
situation has turned this into 
something much, much, much 

225
00:12:43,280 --> 00:12:46,200
larger. 
You know, Chat GPTS daily 

226
00:12:46,200 --> 00:12:52,680
download growth fell by 13% and 
one star review spiked over 775%

227
00:12:52,840 --> 00:12:55,440
on Saturday and doubled again on
Sunday. 

228
00:12:55,920 --> 00:13:01,360
Claw downloads are climbed by 
81% and hit #1 on Apple's US App

229
00:13:01,360 --> 00:13:04,800
Store with over 500,000 
downloads in a single day. 

230
00:13:04,960 --> 00:13:07,080
You know, people aren't just 
debating policy online. 

231
00:13:07,080 --> 00:13:10,000
I think they're they're voting 
with their fingers and their 

232
00:13:10,000 --> 00:13:12,160
funds. 
See, I think there's going to be

233
00:13:12,160 --> 00:13:13,480
a bunch of things to watch for 
here. 

234
00:13:13,840 --> 00:13:16,680
First, whether the supply chain 
risk designation. 

235
00:13:16,680 --> 00:13:19,520
So when the government announced
the cord is a supply chain risk 

236
00:13:20,160 --> 00:13:24,160
actually gets fully formalized. 
It isn't just a tweet so far, 

237
00:13:24,160 --> 00:13:28,400
it's just a social media post. 
It's not a legal proceeding and 

238
00:13:28,400 --> 00:13:31,400
legal experts called it, you 
know, basically more political 

239
00:13:31,400 --> 00:13:33,640
theatre than than an actual show
of force. 

240
00:13:33,720 --> 00:13:37,720
Second, Claude is still the only
model on the Pentagon's 

241
00:13:37,720 --> 00:13:40,800
classified network, and the Wall
Street had reported that it was 

242
00:13:40,800 --> 00:13:44,000
even used in the Iranian strikes
just hours after the ban was 

243
00:13:44,000 --> 00:13:45,880
announced. 
The military would have six 

244
00:13:45,880 --> 00:13:49,320
months to swap the model under 
that directive, but many people 

245
00:13:49,320 --> 00:13:51,400
have already come out in the 
military saying it's going to be

246
00:13:51,400 --> 00:13:54,520
a huge pain in the arse. 
And 3rd, the what could be 

247
00:13:54,520 --> 00:13:55,800
described as the chilling 
effect. 

248
00:13:55,920 --> 00:13:58,920
So, for example, Google has 
quietly removed its long 

249
00:13:58,920 --> 00:14:02,160
standing ban on AI weaponry 
development from its AI 

250
00:14:02,160 --> 00:14:04,640
principles. 
We did that in February 2025 

251
00:14:05,240 --> 00:14:07,840
right before signing an 
agreement with the Pentagon. 

252
00:14:08,520 --> 00:14:10,920
And you know, every tech company
is probably watching what 

253
00:14:10,920 --> 00:14:13,520
happens to your Anthropic, which
is if you step out of line, 

254
00:14:13,880 --> 00:14:37,790
you'll be burnt. 
So yeah, to conclude, I think 

255
00:14:37,950 --> 00:14:39,830
there's been a bit of a crazy 
turn of events. 

256
00:14:39,830 --> 00:14:42,870
I think there's been one of the 
most high profile fallings out 

257
00:14:42,870 --> 00:14:46,320
between AUS company and the US 
government. 

258
00:14:46,400 --> 00:14:48,760
But what's really important I 
think, is that, you know, I 

259
00:14:48,760 --> 00:14:51,320
can't help but coming back to 
the fact that Dario Amadei's 

260
00:14:51,320 --> 00:14:55,400
favorite book, the one that he 
used to buy for every single new

261
00:14:55,400 --> 00:14:58,800
employee and frofic was The 
making of the Atomic Bomb. 

262
00:14:59,640 --> 00:15:02,880
He wanted people at the company 
to understand that what they 

263
00:15:02,880 --> 00:15:07,040
were building could and probably
would eventually become so 

264
00:15:07,040 --> 00:15:10,640
important that the government 
would want to control it. 

265
00:15:11,320 --> 00:15:14,360
And that when that moment 
arrived, the people who built 

266
00:15:14,360 --> 00:15:17,400
the technology would have to 
decide whether they got a say in

267
00:15:17,400 --> 00:15:20,200
how it was used. 
And that moment arrived last 

268
00:15:20,200 --> 00:15:22,480
Thursday. 
One company said no, another 

269
00:15:22,480 --> 00:15:24,480
said yes. 
And the person who wrote the 

270
00:15:24,480 --> 00:15:27,680
memo about doing the right thing
then signed the deal the next 

271
00:15:27,680 --> 00:15:29,200
day. 
I think that says a lot about 

272
00:15:29,200 --> 00:15:31,840
Sam Altman, and it says a lot 
about Open AI. 

273
00:15:31,840 --> 00:15:34,720
And I hope the cancelled ChatGPT
movement continues. 

274
00:15:35,280 --> 00:15:37,760
Anyway, that's it. 
I hope you enjoyed today's 

275
00:15:37,760 --> 00:15:39,720
episodes. 
I hope you learned something and

276
00:15:40,480 --> 00:15:41,880
I'll see you next week.