Latest / Elon Musk Podcast / Anthropic fights the Pentagon

Transcript

Anthropic fights the Pentagon · Elon Musk Podcast

• 0:00
  The Pentagon designated AI company Anthropic a supply chain
• 0:04
  risk and banned it from federal agencies.
• 0:06
  Anthropic refused to lift restrictions on its clawed AI
• 0:10
  regarding mass domestic surveillance and fully
• 0:13
  autonomous weapons. They insisted on human
• 0:15
  oversight. Can the government force a
• 0:17
  private tech company to waive its ethical boundaries during an
• 0:21
  active military conflict? The Department of War, led by
• 0:24
  Secretary Pete Higgseth, used a statute typically reserved for
• 0:27
  foreign adversaries to blacklist Anthropic.
• 0:30
  The law is 10 USC 3-D252. It gives the Defense Secretary
• 0:35
  sweeping authority to exclude sources that pose a national
• 0:38
  security threat. Anthropic is suing the
• 0:41
  administration right now. They claim this is First
• 0:43
  Amendment retaliation and an unlawful secondary boycott.
• 0:46
  This redefines federal procurement completely.
• 0:49
  It changes supply chain risk from a technical threat like
• 0:52
  hardware sabotage into a punishment for ideological
• 0:55
  noncompliance. Which threatens Anthropic's
• 0:57
  commercial contracts. A secondary boycott effects any
• 0:59
  vendor doing business with the military, right?
• 1:01
  If a logistics company sells office supplies to a base, they
• 1:04
  are suddenly banned from using Claude in their private
• 1:07
  accounting department. The Pentagon is actively using
• 1:09
  Anthropics technology to analyze intelligence in an ongoing
• 1:12
  conflict against Iran right now. Operation Epic Fury, they
• 1:16
  deployed the AI to process battlefield data.
• 1:19
  The military argues A vendor cannot veto chain of command
• 1:23
  decisions. This limits the military's
• 1:25
  tolerance for independent corporate ethics.
• 1:27
  It opens up a reality where defense contractors must
• 1:30
  guarantee uninterrupted access to their tools regardless of
• 1:34
  internal policies. You have forces on the ground
• 1:37
  relying on instant data processing.
• 1:40
  A live strike operation cannot wait for an AI company's ethics
• 1:43
  board to clear request. The mission requires absolute
• 1:47
  technical obedience. The government cannot compel a
• 1:49
  domestic company to violate its own safety terms.
• 1:52
  Anthropic negotiated specific limits.
• 1:55
  Forcing compliance under the threat of a federal blacklist is
• 1:57
  executive overreach. Look, the military needs
• 2:00
  absolute reliability in a war zone.
• 2:03
  You cannot run an operation if your software decides a target
• 2:06
  analysis violates its terms of service.
• 2:08
  The military signed a contract knowing exactly what the
• 2:11
  limitations were. They agreed to those terms.
• 2:13
  Just to be clear with you listening, we are just laying
• 2:16
  out the arguments from the source material here.
• 2:18
  We are impartially conveying these details.
• 2:20
  We aren't endorsing the administration's policies.
• 2:23
  We aren't taking the tech sector side either.
• 2:25
  Right. The administration directed
• 2:27
  federal agencies to immediately cease all use of anthropics
• 2:31
  technology. This force defense contractors
• 2:34
  to certify they do not use the company's models anywhere in
• 2:37
  their stack. It limits Anthropics market
• 2:40
  access and forces massive compliance audits across the
• 2:44
  entire defense industrial base. Hours after Anthropics ban, Open
• 2:48
  AI struck a deal to deploy its models on the Pentagon's
• 2:51
  classified networks. Wait back up, how did Open AI
• 2:54
  secure the contract when CEO Sam Altman claims they kept the
• 2:58
  exact same safety restrictions Anthropic was banned for?
• 3:01
  Open AI agreed to an all lawful use standard.
• 3:04
  They implemented technical architecture controls rather
• 3:07
  than anthropics policy based vetoes.
• 3:09
  They're using cloud only deployments and cleared embedded
• 3:12
  engineers. Yeah, this changes the framework
• 3:15
  for military tech partnerships. It proves the government will
• 3:18
  reward technical compliance while punishing policy based
• 3:22
  resistance. Anthropics Claude surged to the
• 3:24
  top of the App Store charts. Users cancelled ChatGPT
• 3:28
  subscriptions in a Quit GPT protest.
• 3:30
  This shifts consumer brand loyalty aggressively.
• 3:33
  It turns AI adoption into a political and ethical statement.
• 3:37
  The Information Technology Industry Council warned against
• 3:40
  the ban. Thousands of workers from
• 3:42
  Google, Amazon and Open AI signed a We Will Not Be Divided
• 3:46
  letters supporting Anthropic. This limits the government's
• 3:49
  ability to quietly isolate Anthropic.
• 3:51
  It opens up a massive rift between elite tech talent and
• 3:54
  government defense initiatives. Open AI raised $110 billion in
• 3:58
  private funding. Amazon invested an initial $15
• 4:01
  billion, which could reach $50 billion total.
• 4:04
  This changes the balance of power and cloud computing.
• 4:07
  It ends Microsoft's exclusive grip on open AI.
• 4:10
  Open AI will use Amazon's Tranium chips.
• 4:12
  They're consuming 2 gigawatts of capacity.
• 4:14
  They're building a stateful runtime environment for AWS
• 4:16
  Bedrock. NVIDIA also reported a 73%
• 4:20
  revenue jump to $120 billion. This ensures massive sustainable
• 4:25
  compute scale for the foreseeable future.
• 4:28
  It limits fears of an AI hardware bubble popping.
• 4:31
  And that stateful runtime environment opens up advanced
• 4:34
  multi step agent workflows for developers.
• 4:37
  If you got a decent mic and laptop and some free time,
• 4:39
  Babble Audio is paying people to record speech data and annotate
• 4:43
  audio for AI training. No minimums, no fixed hours.
• 4:46
  You work when you want and get paid weekly via PayPal, Venmo,
• 4:50
  or bank transfer. They pay per recorded or
• 4:53
  annotated hour plus bonus challenges for hitting weekly
• 4:56
  goals. And if you sign up through our
• 4:58
  link, you get priority processing and a $15 bonus.
• 5:01
  Links in the show notes. A university study found
• 5:03
  developers using AI coding agents are actually working
• 5:05
  longer hours. This shatters the promise of AI
• 5:08
  as a massive time saver. It changes expectations that
• 5:11
  engineers are now pushed into a high pressure race to build
• 5:14
  faster. India's $300 billion outsourcing
• 5:17
  industry employing over 6 million people is facing direct
• 5:21
  automation of of its white collar back office tasks.
• 5:23
  This threatens to collapse the global economic model of
• 5:27
  outsourced labor. It forces entire national
• 5:30
  economies to adapt quickly or face massive job losses.
• 5:33
  A federal judge rejected Anthropics $1.5 billion
• 5:37
  copyright settlement with authors over its use of pirated
• 5:41
  book libraries like Library Genesis.
• 5:43
  The. Pentagon has weaponized this
• 5:44
  ruling. They claim Anthropic rips off
• 5:47
  content and has no right to lecture the military on ethics.
• 5:50
  This limits and tropics moral high ground.
• 5:53
  It shows how corporate legal vulnerabilities can be used
• 5:56
  against companies in geopolitical disputes.
• 5:58
  Data brokers have cost consumers 10s of billions of dollars in
• 6:01
  identity theft losses due to hidden opt out pages.
• 6:04
  This forces a harsh reality check on consumer privacy.
• 6:08
  It opens up new demands for strict federal regulation of the
• 6:10
  data market. Israel hacked a popular Iranian
• 6:13
  prayer app to send push notifications urging defections
• 6:16
  and resistance. This changes the nature of
• 6:18
  psychological warfare. It turns everyday consumer
• 6:21
  religious apps into to direct channels for military
• 6:23
  operations. Let's move.
• 6:24
  Over to. Hardware Xiaomi launched the
• 6:26
  1717 Ultra and Lights Phone in Europe, featuring the Snapdragon
• 6:30
  8 Elite Gen. 5 chipset. This opens up intense high end
• 6:34
  smartphone competition in Western markets against Apple
• 6:37
  and Samsung. NASDAQ partnered with Kraken to
• 6:40
  launch tokenized stocks with 1 to 1 ownership and voting
• 6:43
  rights. This changes traditional market
• 6:45
  infrastructure. It enables continuous 24/7
• 6:48
  trading and bridges legacy finance with crypto.
• 6:51
  LibreOffice version 26.2 added native Markdown support.
• 6:56
  Microsoft 365 introduced a highly expensive E7 Premium AI
• 7:00
  tier. This forces a wedge in the
• 7:02
  software market. It opens up streamlined options
• 7:05
  for open source users while limiting enterprise budgets due
• 7:07
  to Microsoft aggressive AI pricing.
• 7:09
  EV charger company ELECQ suffered A ransomware attack
• 7:13
  exposing customer home addresses.
• 7:15
  Dutch police proactively warned 100 alleged scammers to turn
• 7:18
  themselves in. This changes our understanding
• 7:20
  of infrastructure risk. It proves the green energy
• 7:23
  transition is highly vulnerable to extortion.
• 7:26
  It also opens up new proactive tactics for law enforcement.
• 7:29
  Right. The conflict between Anthropic
• 7:30
  and the Pentagon proves that control over AI is no longer
• 7:33
  just a commercial dispute. It is a fundamental battle over
• 7:36
  national security and corporate independence.
• 7:38
  When AI is integrated into every military and economic system on
• 7:42
  Earth, who actually holds the power to decide where the
• 7:45
  ethical lines are drawn? If you're not subscribed yet,
• 7:48
  take a second and hit follow on whatever app you're using.
• 7:51
  It helps us keep making this. We appreciate you being here.