Nadella calls for decentralized AI governance: Microsoft to release its first model behavior guidelines tomorrow.

As the debate over AI safety sweeps the industry, Microsoft CEO Satya Nadella has announced his participation in the discussion and will be the first to publicly release the behavioral guidelines for his company's AI models tomorrow.

The AI industry is undergoing a rare collective "introspection." Anthropic CEO Dario Amodei published an article last Saturday calling for a "global slowdown in AI development," a sentiment echoed publicly by Elon Musk and Sam Altman. Against this backdrop, on September 14th, Microsoft CEO Satya Nadella posted on the X platform, explicitly supporting "consciously controlling the pace of development" to advance aligned research, and announced that Microsoft will release its first "Code of Conduct" for its MAI model the following day, open for public consultation.

This marks the first time Microsoft has publicly released its behavioral standards for first-party AI models in document form.

Meanwhile, sources familiar with the matter told The Information that Anthropic, OpenAI, and Google are already discussing collaborating to establish an industry standards body for artificial intelligence (AI). OpenAI CEO Sam Altman stated at an employee meeting that he supports the creation of a testing and auditing body for the AI industry.

Nadella draws a red line: AI that is not under human control is "not worth pursuing".

Nadella gets straight to the point in his post: "Any pursuit of superintelligence must be based on a core principle—if the AI we build cannot help humanity or be controlled by humanity, then it is not worth pursuing."

He also emphasized that the benefits of AI must be widely distributed, "benefiting all countries, communities, and businesses," rather than concentrated in the hands of a few players. He believes this requires a cutting-edge ecosystem where both open and closed models can coexist.

At the enterprise level, Nadella's position is equally clear: "Enterprises must have complete control over their own unique, tacit knowledge. Every organization should be able to build its own continuous learning cycle without relying on any single model provider and be able to embed its own knowledge into the models and weights it controls."

Microsoft's three-tiered approach: open access, corporate self-control, and public disclosure of codes of conduct.

In the post, Nadella outlined Microsoft's specific action framework, which is divided into three levels:

First , it provides broad access and choice at every layer of the AI technology stack;

Second , ensure that the enterprise has complete control over its own learning cycle and model;

Third , publish the "Code of Conduct" for the MAI model for public consultation.

He stated that Microsoft welcomes "the research, focus, and deliberate pace required to achieve alignment goals," as well as institutional ideas such as "embedded evaluators," "so that all of this doesn't just remain lip service."

In a blog post last Saturday, Amodei suggested that major AI companies should grant external evaluation agencies, such as Model Evaluation and Threat Research, the same level of system access as their employees. Anthropic has pledged to implement this step, and Altman subsequently stated that OpenAI will follow suit.

Decentralized governance: It cannot be decided by a few entities.

Nadella offered a clear warning about the current AI governance landscape.

He wrote: “The key is that this cannot be controlled by a few entities, but must have broad representation across ecosystems, countries, and sectors, including academia.”

This statement directly echoes the ongoing discussions within the industry regarding the establishment of an industry standards body. According to sources familiar with the matter, Anthropic, OpenAI, and Google formed a working group in July of this year, with representatives from the three companies (below the CEO level) meeting regularly to discuss plans for establishing an industry standards body. Their most recent meeting was last week.

Sam Altman stated at the OpenAI all-hands meeting this week that he supports establishing a testing and auditing body for the AI industry, but believes that "individual labs must push for this themselves and cannot rely on support from the US government," according to a source familiar with the matter. He also stated that there is currently insufficient consensus within the industry, with some companies opposing it, and that "the best approach is to work with other labs to push for it."

White House executive order stalled, industry self-regulation becomes the main theme

Government-level efforts have stalled. As previously reported by The Information, a draft executive order circulated in the White House proposing the establishment of a self-regulatory body for AI, similar to FINRA (Financial Industry Regulatory Authority), but the plan stalled in late summer due to insufficient support within the Trump administration.

Former White House AI director David Sacks has consistently opposed the establishment of a FINRA-like AI regulatory body, arguing that government intervention should be avoided. According to Politico, Meta CEO Mark Zuckerberg also expressed concerns about establishing a national AI regulatory body during an August phone call with Trump.

Sacks stated on the X platform that AI companies can directly advance safety efforts "without needing antitrust exemptions to coordinate." However, he also questioned whether slowing down development was "purely altruistic," pointing out that product liability risks themselves already provide a financial incentive to prevent AI from causing harm.

There has also been criticism from within the industry. Cohere CEO Aidan Gomez posted on the X platform, expressing concern about the largest AI companies coordinating to create rules that could restrict smaller competitors, writing directly: "Here comes a great idea from the cartel."

Risk Warning and DisclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.