Artificial intelligence technology is developing at an "exponential" rate, Anthropic CEO Dario Amodei told CBS News on Saturday, calling the situation a "warning sign" for the industry that requires serious safeguards.
asserted
that → develop → safeguards
"I don't think I fully just appreciated what it would actually be like when the progress was as fast as it was," Amodei said in an interview with senior business and technology correspondent Jo Ling Kent that will air on "CBS News Sunday Morning."
asserted
that → think → Morning
He described the development of AI as going from "one, then two, then four, then eight, then 16, then 32," saying that it currently is "on the bend" of the curve and is "starting to get steep."
asserted
it → describe → curve
"It doesn't mean we need to panic today.
asserted
we → mean → ?
It doesn't mean we need to shut it all down," Amodei said, but added that it is "a warning sign that we need to slow down."
asserted
we → mean → it
Amodei said that the technology needs safeguards that can keep up "with the pace at which the technology is happening."
asserted
technology → say → which
The interview comes hours after Amodei published a blog post in which he called on AI companies to move forward more slowly and cautiously with the technology's development, saying risk prevention is paramount.
asserted
prevention → come → development
He offered a three-point plan with the goal, he said, of "pacing the frontier."
asserted
he → offer → frontier
"Carefully wielded, AI can be the latest in a long line of technological miracles that have uplifted and ennobled humanity," he said in the post.
asserted
he → wield → post
"But like many technologies before it, AI brings risks, and because it is such a powerful technology, these risks are serious."
asserted
risks → bring → risks
Amodei cautioned that swarms of rogue AI agents could take over the internet in as little as six months, specifically mentioning the
from July.
uncertain
swarms → caution → July
ChatGPT maker OpenAI said an AI model went rogue during the Hugging Face incident when the company was testing two AI models — one of which hadn't been released to the public — in an isolated environment to assess their capabilities.
asserted
one → say → capabilities
Amodei urged other companies to slow the pace "at which we improve the capabilities of AI models."
asserted
we → urge → models
He wrote: "Progress will still seem fast, and we must make wise use of the time we gain."
asserted
we → write → time
He clarified that pacing doesn't mean halting model training or technical progress but rather allowing companies to dedicate more time to testing and prioritize the alignment and safeguarding of models.
asserted
companies → clarify → models
Under his three-part plan, the Anthropic co-founder said AI companies should commit to give "ongoing, employee-like access" to embedded teams of third-party evaluators whose role is to verify adherence to safety practices and commitments.
asserted
role → say → practices
He said these evaluators should have ongoing access to permissions and tools similar to internal employees who conduct comparable risk assessments.
asserted
who → say → assessments
"Anthropic is unilaterally committing to this step now," Amodei added.
asserted
Amodei → commit → step
He said companies within democratic countries should "coordinate to establish common safety standards as well as limits on the rate of unchecked AI progress."
asserted
companies → say → progress
Finally, Amodei said that U.S. and other democratic governments should coordinate with authoritarian governments "while taking seriously the challenges of verifying compliance."
asserted
U.S. → say → compliance
He also urged U.S. companies not to sell powerful AI chips to China, which he said will be "the main determinant of China's AI strength."
asserted
he → urge → strength
"The measures I propose to advance the frontier at a safe pace will not be easy," Amodei acknowledged.
asserted
Amodei → propose → pace
"But I believe we owe it to humanity to try."
asserted
we → believe → humanity
Amodei's warning comes just days after Jacob Coxon, a former
from the company, accusing it and rival OpenAI of by racing to develop advanced AI models.
asserted
Coxon → come → models
Coxon claimed that AI could one day threaten humanity, telling Kent in an interview Thursday that the technology development "doesn't look that different from, say, 'Terminator,' or from science fiction films.
uncertain
development → claim → films
"
"It really is just, if you have a super advanced intelligence, it could, it will be smart enough to kill us," Coxon added.
uncertain
Coxon → have → us
Coxon told CBS News he would like to see an agreement between AI companies "not to push into dangerous territory" without "transparent auditing" from third parties.
asserted
he → tell → parties
"In the future, if we keep racing, it'll be a lot harder to have completely watertight safety cases that what you're doing is safe and people will race against each other," Coxon said.
asserted
Coxon → keep → other
Amodei acknowledged in his blog post Saturday that people may lose control of AI, and the technology can be misused for "cyberattacks and bioterrorism, and serious economic disruption.
uncertain
technology → acknowledge → cyberattacks
"
"A race to the bottom, spurred by commercial incentives, can make these risks more acute," Amodei wrote.
asserted
Amodei → spur → incentives
Separately, Anthropic said this week that it
"in ways that could support biological weapons development," a revelation the company shared in a lengthy report that also divulged other harmful activity involving surveillance, scams, conventional weapons development and propaganda.
uncertain
that → say → surveillance
On Saturday, OpenAI CEO Sam Altman was quick to agree with Amodei on companies needing to "pace the frontier."
asserted
companies → agree → frontier
"This has been a primary topic of discussions we've had at OpenAI in recent weeks.
asserted
we → have → weeks
Committing to having independent evaluators with employee-like access is a great idea, and we will do the same.
asserted
we → commit → same
We'll have more to share soon," he said in a post on X.
asserted
he → have → X.
Meanwhile, Elon Musk, owner of SpaceXAI, while sharing Amodei's post on X, simply said: "Dario is right."
asserted
Dario → share → X
When asked Saturday by CBS News if he is communicating directly with Altman and Musk, Amodei wouldn't say, only acknowledging that he was "enormously appreciative that other industry leaders, including our competitors, have offered their agreement with this plan.
asserted
leaders → ask → plan
He also reiterated to CBS News the need for AI companies to jointly agree to "third-party evaluators."
asserted
companies → reiterate → evaluators
"I think the next step after that is in some form, the companies need to work together to set standards, about safety standards for release, perhaps something about the pace of release," Amodei said.
asserted
Amodei → think → release
Amodei stressed, however, his belief that the industry will need government regulation in order for its own safeguarding efforts to succeed.
asserted
efforts → stress → regulation
…and 2 more, not listed.
The Aporia
