Anthropic Co-founder and CEO Dario Amodei: The Industry Has Been Lying for Too Long
Anthropic co-founder and CEO Dario Amodei stated in an interview on CBS's "Sunday Morning" aired on September 13, "I think this industry has been lying to the public about AI risks for a long time." The interview was recorded at Anthropic's San Francisco headquarters and hosted by CBS News reporter Jo Ling Kent, just a day after he published an article calling for a slowdown in AI development.
In the interview, Amodei acknowledged that "there are indeed real dangers" and expressed regret for not fully realizing the rapid pace of AI advancements earlier. He specifically proposed policies including that AI companies should "slow down" their development pace, introduce third-party evaluators to test new models (comparing these evaluators to food safety inspectors), and expressed support for implementing a "kill switch" for AI systems, stating that this "might be a good idea."
However, he clearly opposed a complete ban on AI development, citing concerns that authoritarian countries would continue to advance this technology unchecked, putting democratic nations at a technological disadvantage. He advocated for "oversight by democratic governments" and "joint governance," rather than relying solely on self-regulation by private companies, admitting that he feels uneasy about "AI technology development being entirely led by private companies."
The report mentioned two related legislative proposals: the "Artificial Superintelligence Ban" bill proposed by Senators Sanders and Casar, and the AI "kill switch" bill proposed by Representatives Lieu and Moran. Amodei's statements were interpreted as more favorable towards the latter and more reserved regarding the former.
The interview also noted that former Anthropic researcher Jacob Coxon had publicly warned the previous Tuesday about the potential emergence of "superhuman systems that could infiltrate all systems," which corroborates Amodei's assertion in his article that "AI capabilities have accelerated dramatically since this summer." According to Forbes, Amodei's net worth is approximately $15.5 billion, ranking him 193rd on the global billionaire list. Following his statements, other tech billionaires quickly expressed support for calls to slow down development.
Market-wise, this round of statements comes at a delicate moment as Anthropic is racing towards a Nasdaq IPO (with market valuation expectations once inflated to $2 trillion) while the CEO publicly admits that "the industry has been lying" and calls for a slowdown. Sending out safety signals helps Anthropic to establish a public image as a "responsible AI company" before regulatory frameworks take shape, adding positive leverage to its IPO narrative. However, such statements could also be interpreted by capital markets as signals of uncertainty, especially in light of previous analyst warnings about "AI hyperscale cloud vendors' free cash flow turning negative by 2027." If the "slowdown" rhetoric is over-interpreted, it may further amplify market caution regarding valuations in the AI sector.
Source: Public Information
ABAB AI Insight
Amodei's "industry lying" remarks are not isolated—he and his sister Daniela Amodei left OpenAI in 2021 due to disagreements over safety issues, founding Anthropic with a core narrative of "AI safety" as a competitive differentiator. His public acknowledgment of the "industry lying for a long time" actually targets the widespread practice in the industry, including Anthropic itself, of "shouting safety slogans while frantically expanding computing power" over the past few years.
A notable timing aspect is that this interview coincides with Anthropic's critical window for its Nasdaq IPO—having recently completed a $65 billion Series H funding round, reaching a valuation of $965 billion, with market expectations for the IPO valuation potentially hitting $2 trillion. At this juncture of financing and listing, the CEO's proactive role as an "industry whistleblower" can objectively help the company accumulate brand assets of "willingness to self-criticize and be trustworthy," essentially trading safety discourse for regulatory benefits and public trust, aiming for a more favorable narrative environment for the upcoming public market pricing.
This "founder publicly criticizing the industry's chaos while also being part of it" stance is similar to the "confessional statements" made by executives of several leading companies in history, such as some Wall Street executives reflecting on excessive leverage after the financial crisis or social media executives admitting to addictive design issues. However, Amodei's criticism occurs while the industry's capabilities are still rapidly rising and the regulatory framework is not yet defined, rather than as a post-factum regret, making his statements closer to "laying the groundwork for potential regulatory tightening" during a critical window of capability expansion and safety narrative tension.
Essentially, this is a power struggle for discourse ahead of regulatory changes—when legislation (like the Sanders-Casar superintelligence ban and the Lieu-Moran kill switch bill) has already entered the congressional agenda, leading companies admitting past issues and proposing "moderate regulatory solutions" (third-party evaluations + kill switches, rather than a complete ban) is a mechanism to predefine the boundaries of "reasonable regulation" and avoid more radical legislative proposals as the final outcome. This "self-criticism for agenda-setting power" strategy effectively shifts some regulatory discourse power from legislators back to the regulated entities themselves.
ABAB News · Cognitive Laws
- Those who admit mistakes first have the right to define the boundaries of punishment.
- The louder the safety narrative, the closer it is to the regulatory window.
- The larger the financing, the more a self-critical image is needed to back it up.