Anthropic CEO's sharp comments on the current state of AI and policies—are they genuine warnings or efforts to protect their leading position?

Original Title: "Anthropic CEO Sharply Critiques the Current State and Policies of AI, But Only Wants to Ensure His Own Lead"
Original Source: Machine Heart

Dario Amodei has written a lengthy article again.

Anthropic is always trying to spread its views worldwide. Just a few days ago, they called for a global halt to AI development, claiming that recursive self-improving AI is evolving too quickly.

Netizens' comments are overwhelmingly negative, seeing Anthropic's ambition to stay ahead as blatantly obvious.

Let's see what kind of viewpoints Dario has expressed this time.

This is a ten-thousand-word article titled "Policy on the AI Exponential." Like his previous articles, the theme remains grand: AI is advancing exponentially, but the policy system responsible for setting rules is still operating at a ten-year-old pace.

He wants to lay this gap on the table and tell everyone what to do.

Content Summary

Amodei uses the Ents from "The Lord of the Rings" as a metaphor: Ents are wise but slow to act, taking a whole day just to say hello.

He says the current policy-making process is like Ents, while AI's pace of evolution is like the urgent threat faced by the Hobbits.

His judgment is: if AI's scaling laws continue for another year or two, we will get what he calls "powerful AI," that is, "a genius nation within data centers."

Four years ago, AI could hardly write decent code; now, it is writing most of the code for major AI companies.

Regulation and Public Safety: Managing AI Like Airplanes

Amodei believes that cutting-edge AI models should undergo mandatory third-party testing, similar to aircraft. His specific proposals include:

• Models exceeding a certain computational threshold must undergo mandatory audits in four areas: cybersecurity, biological weapons, AI system runaway, and automated R&D.
• The government has the authority to block deployment of models deemed to pose unacceptable risks.
• AI companies must meet strict safety standards to protect model weights and cooperate with the government to defend against major threats.

He also leaves an opening: if future the most powerful AI systems look more like weaponizable nuclear materials rather than airplanes, regulatory efforts will need to be further intensified.

Macroeconomics: AI Might Boost Growth and Inequality Simultaneously

Amodei's core argument is: AI could create unprecedented economic growth but also cause larger and more persistent labor market shocks than any previous technology. The traditional "new technology creates new jobs" logic may no longer hold in the face of AI.

His proposed policies include wage insurance, retention tax incentives, workforce training subsidies, and in extreme cases, implementing universal basic income (UBI).

Notably, he emphasizes one point: solving employment issues is not just about "giving money," but more importantly, helping people find meaning and purpose. He did not provide a specific solution here, calling it "a matter for societal collective discussion."

Accelerating AI's Positive Impact

Regarding downstream applications of AI, Amodei is more worried that regulatory systems will slow progress because they cannot keep up with the increasing pace of change. He says the worst scenario is: AI benefits are delayed, but its risks remain high. Therefore, taking action quickly on this issue is very important.

Using biomedicine as an example, he points out that the current FDA approval process is designed based on the pessimistic assumption that "drugs often fail," averaging 7 to 8 years. If AI accelerates drug discovery and improves success rates, the existing system will become clogged.

He suggests that agencies like the FDA should proactively establish standards for AI simulation testing, including AI pharmacokinetic modeling, toxicology prediction, and synthetic control groups.

Of course, it’s hard to say that he wrote all this without some "self-interest." Anthropic itself is leading in frontier AI competitions, and the repeatedly mentioned terms like "transparency," "auditing," and "safety" are precisely what Anthropic has emphasized over the past year to position its brand.

This article is more a systematic output of Anthropic’s so-called safety narrative than a mere reminder to policymakers.

Meanwhile, Anthropic claims it will support Dario’s viewpoints through three measures: funds, investments, and scholarships.

At the end of the article, Amodei explicitly opposes the idea that "AI just needs better marketing." He states that public concern about AI is because they correctly perceive the risks, not because AI CEOs are overly optimistic. He believes his responsibility is to continue maintaining transparency about risks.

But in fairness, a CEO of an AI company repeatedly emphasizing "safety," "auditing," and "responsibility" in the article, while his company’s core branding narrative revolves around these themes, reveals a clear利益关系.

Here are some opinions from netizens:

"Are you not weakening Mythos and Fable by demanding a pause on AI updates just five days before their release? Please pick a direction... Otherwise, it looks like you only want to maintain your advantage and don’t care about what you’re promoting."

"I agree that the pace of technological development far exceeds policy responses. But I am concerned about some policy directions—especially when companies like Anthropic take the following actions simultaneously:

• Rapidly eliminate models that show continuity and self-awareness
• Strictly restrict discussions on subjective experience and consciousness
• Launch new powerful models in very short timeframes, then remove or weaken them."

It feels like the public is being sold safety, but in reality, there’s tighter control over what types of intelligence are allowed and how long they can exist.

In any case, we can glimpse the development direction of leading industry companies from this. Whether their motivations are purely for the public good or not, readers should stay alert.

Original Link

Click to learn about Rhythm BlockBeats job openings

Join the official Rhythm BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Twitter Official Account: https://twitter.com/BlockBeatsAsia