Revolt against AI: How Corporations Profit from Neurophobia
The heads of the largest American corporations specializing in the development of artificial intelligence (AI) — OpenAI, Anthropic and SpaceX — simultaneously called for a slowdown in its development. Corporations have drawn attention to the apocalyptic threats associated with the rapid progress in this area. However, it seems that the main motivator for AI leaders' bosses is not concern for the fate of humanity, but a much more mundane material interest. "Izvestia" investigated why the largest market players verbally oppose the development of this market and what their benefits might be.
Bees versus honey
Few people are surprised by the alarming statements from the head of Anthropic, Dario Amodei, about the threats of the rapid rise of AI. He took this position a year or two ago. What is much more interesting is that this time the leaders of OpenAI and Google DeepMind agreed with him. And what is quite surprising is that SpaceX CEO Elon Musk, who has always had a "special opinion" in this industry, also supported him. Even investors were impressed by this unanimity. For a short period of time, shares of the leading companies in the sector, which are breaking records in terms of market value (especially taking into account meager profits or, conversely, serious losses), even lost value during trading on NASDAQ.
In fact, the calculation here is quite subtle and obvious. The main goal of the "slowdown" campaign is to force the government, primarily the American government, to introduce strict regulation of the industry. If the government legislatively recognizes AI as a deadly technology, the next step will be to introduce barriers.
Among the proposed measures will inevitably be mandatory government licensing of data centers, requirements for conducting security audits before each release, and the introduction of strict legal liability for any unforeseen actions of neural networks.
For corporations like OpenAI (backed by Microsoft's bottomless budgets) or Anthropic (subsidized by Amazon and Google), spending hundreds of millions of dollars on lawyers, auditors, and compliance departments is a routine transaction cost. For small independent startups, academic labs, and open source projects, such requirements will become deadly. We are talking not only about independent developers, but also about Chinese companies, about the dangers of the open-source model of which the American and European press has written quite a lot in recent months. Complex bureaucratic procedures negate all their modern advantages. Ideally, of course, it is planned not to be limited only to the United States, but through compliance procedures to extend deterrence to all countries that trade and interact with America a lot.
Uncomfortable questions
Discussions about AI destroying humanity in five or 10 years are beneficial to corporations for a variety of other reasons. For example, to divert the attention of congressmen and regulators from other countries from the quite mundane legal problems of bigtech.
High-tech giants regularly have to answer uncomfortable questions. On what basis were algorithms trained on copyrighted materials (including multibillion-dollar lawsuits from The New York Times, writers and artists guilds)? How is the problem of unauthorized use of personal data of hundreds of millions of people solved? What should we do about monopolizing the computing power market? Discussing the "risk of a machine uprising" at a hearing in the US Senate is much safer for capitalization than answering legally tricky questions about intellectual property issues and ensuring that billions of dollars in compensation are paid to copyright holders.
Money talks
The position of AI bosses looks particularly ambiguous when comparing their public statements with real corporate strategies. None of the mentioned companies stopped the purchase of chips and the training of the next generation of models for a single day. On the contrary, Sam Altman, the head of OpenAI, continues to look for trillions of dollars in the Middle East and Asia to build giant clusters.
The story of the California bill SB 1047 is typical. In 2024, state politicians tried to move from abstract conversations to business, proposing to introduce direct legal liability for developers for catastrophic damage from their algorithms. OpenAI officially opposed this document by connecting its entire lobbying resource. As soon as the government tried to introduce real mechanisms of punishment for businesses, the enthusiasm of corporations in the field of "security" instantly evaporated.
The licensing initiative is fraught with another legal trick. By giving the state the right to issue permits for launching models, corporations share responsibility with it.
If in the future a government-certified neural network is used by hackers to write malicious code or hack into financial infrastructure, developers will get an ideal argument to defend themselves in court. They will be able to claim that the model has passed all the tests established by the state, has received a license, and therefore the fault lies with the regulator itself. Strict control in this case acts as an insurance policy and an indulgence from multibillion-dollar class action lawsuits. A powerful PR effect is added to this: companies position themselves as responsible visionaries who are ready to sacrifice profits for the good of the planet.
This week, analyst Kevin Bass conducted an audit of the finances of Anthropic and found out that the "independent" organization METR, which, according to corporate leaders, should check the company's products for security, is a major investor in AMD. The idea looks perfect: whether AI from one company or another can destroy humanity will be judged by the de facto organization under their control. "Everything is for our own people, for strangers it is the law."
Reactions and perspectives
Predictably, the idea of artificially slowing down the American technology industry has met with sharp opposition from the current administration. US President Donald Trump has described calls for a slowdown as outright sabotage. From the point of view of the conservative establishment, the development of AI is an arms race. Any regulations that restrict American business are perceived as direct sabotage of national security.
The fears of American politicians have a reinforced concrete foundation: Beijing is not going to slow down. Chinese corporations (Baidu, Alibaba, Tencent) have made a powerful leap in recent months, flooding the global market with high-quality and cheap open source solutions. For China, talking about a pause in the United States sounds like an invitation to take a leading position in the markets of the Global South. The attempt by American corporations to build regulatory barriers within the United States will lead to the fact that the center of technology development will finally shift towards Asia.
The economic model of modern bigtech eliminates the possibility of a real pause. The industry operates at the expense of huge investments. This year alone, hundreds of billions of dollars have been allocated for the purchase of equipment and the construction of data centers.
Investors estimate the capitalization of the US technology sector at trillions, based on expectations of a non-stop, exponential growth of neural network capabilities. A real physical shutdown of development for at least six months will cause a domino effect. The market is aware that new products will not appear on time, and subscription revenue will not cover the cost of electricity and chips. It is also worth considering that AI and related industries actually support the entire American stock market. If they start a recession, everything else will just fall apart.
That is why no real server shutdowns are expected. Letters warning of the apocalypse will continue to be published, and corporate leaders will call for the creation of commissions and committees. Under the guise of these discussions, AI majors will continue to reshape the global market, trying to set the most favorable rules for themselves.
Переведено сервисом «Яндекс Переводчик»
