Anthropic CEO Dario Amodei unveiled a three‑step industry safety framework over the weekend, sparking a clear divide among the United States’ most influential AI leaders.
The three‑step proposal
Amodei’s plan calls for (1) embedding third‑party evaluators inside AI labs, (2) coordinating safety efforts across the domestic AI industry, and (3) forging international agreements that could involve government assistance. The Verge reported the wording verbatim: “Over the weekend, Anthropic CEO Dario Amodei had proposed a three‑step plan for slowing AI development, including by embedding third‑party evaluators in labs, coordinating across the domestic industry, and forging international agreements potentially with government assistance.”
The proposal is framed as a voluntary industry response that would give external experts direct access to model training data and code, a step many observers have said could improve transparency without imposing a new regulatory body.
Support from OpenAI and SpaceX
OpenAI’s Sam Altman appeared to back key elements of the plan. The Verge noted that “OpenAI CEO Sam Altman, Google DeepMind co‑founder Demis Hassabis, and even SpaceX CEO Elon Musk publicly seemed to agree on aspects of all three things.” While Altman did not issue a formal endorsement, his public comments aligned with Amodei’s emphasis on third‑party evaluation.
SpaceX CEO Elon Musk also signaled agreement with the safety steps. The same source excerpt groups Musk with Altman and Hassabis, indicating Musk’s willingness to consider third‑party evaluators and industry coordination.
OpenAI’s global affairs chief Chris Lehane reinforced the notion that company‑led safety actions are essential, telling The Verge, “People want to know AI is being developed safely… That begins with the steps companies like ours take on our own, but government has an important role too.”
Rejection from Meta, Nvidia and a regulator showdown
By contrast, Meta’s Mark Zuckerberg and Nvidia’s Jensen Huang publicly opposed an industry‑funded independent regulator modeled after the Financial Industry Regulatory Authority (FINRA). The Verge cited a Wall Street Journal report that “Zuckerberg, Musk, and Nvidia CEO Jensen Huang scuttled proposals for an industry‑funded independent regulator for the AI industry — akin to the Financial Industry Regulatory Authority (FINRA).”
Zuckerberg’s stance was explicit: “Meta CEO Mark Zuckerberg quickly came out against limiting companies’ autonomy… ‘Every lab has the responsibility and incentive to move at the pace required to train its models safely…’” The quote shows his belief that voluntary safety measures, not external oversight, should drive the industry.
Jensen Huang’s opposition was reported in the same WSJ summary, indicating that Nvidia’s leadership shares Zuckerberg’s view that a regulator would constrain innovation.
Why the split matters
The divergence highlights a broader strategic question: should AI safety be governed by a self‑policing industry coalition, or by a new regulator with enforcement powers? Proponents of the three‑step plan argue that third‑party evaluators can provide credible oversight without the bureaucracy of a FINRA‑style body. Opponents fear that an industry‑funded regulator could become a de‑facto licensing authority, limiting the speed of model development and potentially harming U.S. competitiveness.
The Wall Street Journal’s reporting, as relayed by The Verge, suggests that the regulator proposal was actively blocked earlier this week, meaning that any future push for such a body will face coordinated resistance from the sector’s most powerful CEOs.
Financial backdrop: Meta’s latest filing
Meta’s most recent Form 10‑Q, filed July 30 2026, shows a net income of $42.621 billion for the six‑month period ending June 30 2026, up from earlier quarters. The filing also lists total assets of $449.956 billion and shareholders’ equity of $261.221 billion as of the same date. While these figures do not directly relate to the AI safety debate, they illustrate the scale of the companies whose CEOs are shaping policy.
Meta’s 2018 revenue of $38.924 billion, reported in a 10‑Q filed October 31 2018, provides a historical reference point for the company’s growth trajectory. The contrast between a 2018 revenue figure and a 2026 net‑income figure underscores the firm’s expanding financial footprint, which may influence its appetite for regulatory constraints.
CEO positions at a glance
| CEO | Company | Support for three‑step framework | Opposition to industry‑funded regulator |
|---|---|---|---|
| Dario Amodei | Anthropic | Yes – proposer | No public comment |
| Sam Altman | OpenAI | Yes – publicly seemed to agree | No public comment |
| Elon Musk | SpaceX | Yes – publicly seemed to agree | Yes – helped scuttle regulator proposal |
| Mark Zuckerberg | Meta | No – emphasized autonomy | Yes – publicly rejected regulator |
| Jensen Huang | Nvidia | No – no comment on framework | Yes – publicly rejected regulator |
Source: The Verge – “The AI regulation smackdown isn’t over” (Sept 19 2026), which cites the Wall Street Journal and direct CEO remarks.
What’s next?
The next weeks will test whether the three‑step plan can gain traction without a formal regulator. If Anthropic and its allies can demonstrate measurable safety improvements, they may persuade reluctant firms to adopt the evaluators voluntarily. Conversely, if the opposition succeeds in keeping a regulator off the table, policymakers may look to Congress for a top‑down solution.
For investors, the split signals that AI‑related governance risk is becoming a material factor in valuation models. Companies that embrace third‑party evaluation could see a premium for perceived safety, while those that resist may face heightened scrutiny from regulators and the public alike.
Until a clear regulatory path emerges, the industry will continue to navigate a “CEO‑against‑CEO” landscape, with each leader’s public statements shaping the debate as much as the technology itself.