Latest / Best of: Features / Anthropic's AI panic has finally reached Congress

Transcript

Anthropic's AI panic has finally reached Congress · Best of: Features

• 0:05
  Anthropics AI panic has finally reached Congress. This
• 0:12
  article is being read by an automated voice generated by Everlett. California
• 0:17
  lawmakers are demanding Congress immediately pass artificial intelligence safety laws as panic
• 0:23
  grips much of the tech industry over fears of AI leading to
• 0:26
  human extinction. Reps. Sam Licardo and George Whiteside's Democrats from California
• 0:32
  co signed a letter late Friday to House Speaker Mike Johnson, urging
• 0:36
  the Republican leader to take immediate action on The
• 0:38
  catastrophic risk that the new technology poses, and requesting that Congress not
• 0:43
  leave the Capitol until a new AI safety measure is passed. The
• 0:47
  letter came just days after an anthropic safety researcher resigned from the
• 0:51
  San Francisco Company and warned that there's a ten percent chance AI
• 0:55
  could kill all humans within the next decade. The disclosure set off
• 0:59
  a wave of panic across the country. President Donald Trump and Johnson
• 1:03
  downplayed those fears over the weekend and emphasized that if American regulators
• 1:08
  slow down AI development, the US could risk falling behind China. We
• 1:13
  cannot put a moratorium on this because China will overlap us, Johnson
• 1:17
  told CNN anchor Jake Tapper on Sunday. He called on the AI
• 1:20
  companies to instead establish guardrails themselves. Them problem is they're all competitors.
• 1:26
  Each one of them wants to maintain an edge over the others,
• 1:29
  he said on CNN. We need to get them all in a
• 1:32
  room at the same time and agree on what the guardrails are.
• 1:35
  That's the only way to do this. Dario Amode, the CEO of
• 1:39
  Anthropic, validated some of his former staffers concerns in a letter published
• 1:44
  Saturday that called for the industry to slow down on development to
• 1:48
  allow room for safety measures, including permanent third party observers inside the
• 1:52
  companies and new government safety standards. Amode stressed that without such guardrails,
• 1:57
  the Internet could be taken over by rogue agents within a year.
• 2:01
  Elon Musk, who SpaceX is a leader in AI, wrote in a
• 2:04
  post on X that Dario is right. OpenAI CEO Sam Altman also
• 2:08
  expressed support Saturday for Amade's essay and announced OpenAI will delay its
• 2:12
  public listing over AI safety concerns. The House, which plans to take
• 2:16
  the last two weeks of September and the entire month of October
• 2:20
  off, is already considering multiple AI safety bills. Representative Ted Liu, a
• 2:26
  California Democrat, proposed a bill that would impose a kill switch to
• 2:29
  AI, aiming to create a way to immediately shut down an AI
• 2:33
  model if it becomes dangerous. Representative J. Obernolte, a California Republican, has
• 2:38
  proposed creating a national framework for how to regulate The development and
• 2:42
  deployment of new technology models. Independent Vermont Senator Bernie Sanders has called
• 2:48
  on Congress to go even further, introducing a bill that would pause
• 2:51
  all AI development and prevent any AI intelligence from surpassing that of
• 2:56
  human intelligence. The first version of the bill proposed up to twenty
• 3:00
  years of prison time if anyone violates the law. Congress already has
• 3:04
  plans to vote this week on a bill that would stop data
• 3:07
  centers from increasing Americans energy costs. Not all tech leaders agree with
• 3:12
  the AI panic. Jensen Huang, the CEO of NVIDIA, which sells many
• 3:17
  of the computer chips that power the AI race, has been more
• 3:20
  skeptical that the AI models will kill all of humanity. He downplayed
• 3:24
  The anxieties at a Goldman Sachs conference on Thursday, and suggested that
• 3:27
  the cybersecurity industry has financial incentives to create fear about online safety.
• 3:33
  And what better way to create demand than to create a problem?
• 3:36
  He said. Who doesn't want their market to be hysterical about their
• 3:40
  product, and line up around the corner for it? The Trump administration
• 3:44
  does not appear interested in moving fast on AI regulation. In May,
• 3:49
  Trump’s postponed an executive order that would have created a framework for
• 3:53
  AI developers to disclose to the government any new developments before coming
• 3:57
  public with them, as he said it would compromise the country's competitiveness.
• 4:02
  He later signed a watered-down executive order that asks AI companies to
• 4:06
  voluntarily submit their products to national security risk reviews. David Sachs, Trump's
• 4:12
  AI czar who has personally invested in at least four hundred and
• 4:15
  forty nine AI related companies, wrote a lengthy ex post on Saturday
• 4:20
  accusing Amode and Altmann of being self interested in their calls for
• 4:24
  a slowdown and telling them to just pause their AI development without
• 4:27
  government intervention. Trump wrote on Truth Social on Monday morning that Amadei
• 4:32
  is pretending to be a perfect little angel, and likened the fears
• 4:35
  around AI to a conspiracy. Vice President J. D. Vance called it
• 4:40
  weird that tech leaders are asking Congress to regulate them. He was
• 4:44
  skeptical that the move is a bit of a Trojan horse.