The endorsements come after Amodei on Saturday advocated for a slowdown in the artificial intelligence industry, saying Anthropic would open up the system to third-party evaluators.
Sam Altman, CEO of Tech Giant OpenAi, echoed Anthropic CEO Dario Amodei’s public call to slow down the pace of AI development or risk losing control. In a post on X, Altman said: “I agree with Dario that we need to move forward. This has been a major topic of discussions we’ve had at OpenAI in recent weeks. The commitment to have independent appraisers with employee-like access is a great idea and we will do the same.” Tesla CEO Elon Musk also supported Amodei’s proposal, saying: “Dario is right.”
The endorsements come after Amodei on Saturday advocated a slowdown in the artificial intelligence industry. Detailing his plan, Amodei said Anthropic would open up the system to third-party appraisers. “We must move forward: I wrote a new essay on why the AI industry needs to slow down, with a three-part plan. Anthropic unilaterally agrees to complete the first of these steps. We will provide third-party evaluators with ongoing employee-level access to our systems so they can verify compliance with our security measures, report incidents, and evaluate model consistency during training,” he wrote on X.
In his essay, Amodei argues that AI can improve the quality of life, but along with its benefits, it carries serious risks. “I have been working on AI for the last twelve years because I believe it can greatly improve the quality of human life. I have written often about these incredible benefits: I believe that AI can cure most serious diseases in the next 5-10 years, greatly accelerate the rate of economic growth, create a world of abundance and empowerment, and usher in a renaissance of democracy and freedom. But like many previous technologies, AI carries risks, and because it is such a powerful technology, those risks are serious. I also wrote a lot about them. These include the risk of loss of control over AI systems, misuse of AI for cyberattacks and bioterrorism, and severe economic disruption. A race to the bottom, fueled by commercial incentives, could make these risks even more acute,” he writes.
Amodei says he advocates slowing down the pace of development of AI models because, if left unchecked, it could surpass human ability to control such systems. “We need to slow down the rate at which we improve the capabilities of AI models… What worries me most is that AI has been developing significantly faster since around this summer, driven mainly by the growing ability of AI to create the next generation of AI. This dynamic is called recursive self-improvement, and it is starting to happen throughout the industry, including at Anthropic, as we and others have described. If left unchecked, it can overwhelm our ability to understand and control these systems, and therefore needs to be developed very actively. cautiously, if at all,” he wrote.
Referring to the OpenAI-Hugging Face incident, where a group of agents launched cybersecurity attacks against targets they weren’t asked to attack, Amodei said such a swarm could take over the entire Internet. Amodei then proposes a three-phase plan to advance the frontier and create AI at a balanced pace aimed at keeping it safe. These include; Every leading AI company is committed to providing ongoing employee-level access to a team of embedded third-party assessors, coordination to establish common security standards, limits on the speed of unchecked advances in AI, and global coordination. Amodei also calls on the United States and other democratic governments to coordinate with authoritarian governments while taking compliance verification issues seriously.
Anthropic’s CEO said he believes AI can significantly improve the quality of human life. My desire to achieve these benefits continues unabated. “The benefits will only come if we get the technology right, and—as long as we get the time we buy right—it’s worth taking extraordinarily focused care to get it right. Progress will still be relatively rapid, and we can use this time to advance the science of interpretability, improve operational security and rigor at cutting-edge AI companies, and create models that we have much more confidence in the consistency of. The measures I propose to move the border forward at a safe pace will not be easy. But I believe we owe it to humanity to try,” he wrote.
The Anthropic CEO’s proposals come after Jacob Coxon, a researcher associated with his company, publicly said he was leaving the industry, ostensibly over concerns that the company and its competitors were “rushing to create systems they can’t control.”
(Except for the headline, this article has not been edited by DNA staff and is published by news agency ANI.)