Anthropic's Amodei leads AI safety camp as Nvidia's Huang pushes back

The global argument over artificial intelligence has hardened into three camps, and the companies building the technology are now the ones arguing loudest about how dangerous it is.

As eNCA reported on 15 September, the debate has split between safety-focused researchers, technology accelerationists, and large corporations trying to avoid being caught on the wrong side of the political fight. The dividing lines run through the boards of the companies themselves.

The safety camp is led by an AI company

Dario Amodei co-founded Anthropic, now one of the largest artificial intelligence companies in the world and heading toward a public listing. He left OpenAI because he believed the company led by Sam Altman was not taking the technology’s potential and its dangers seriously enough.

That position draws criticism from rivals who call it hypocritical, since Anthropic is itself pushing out cutting-edge systems.

Geoffrey Hinton, a former Google researcher, and Yoshua Bengio led warnings about AI’s threats in the months after ChatGPT was released in 2022. Max Tegmark of the Massachusetts Institute of Technology, who heads the Future of Life Institute, organised an open letter in 2023 calling for a six-month pause in developing the most advanced systems. Thousands of people signed it, including Elon Musk.

Further outside the industry sits Eliezer Yudkowsky, a self-taught theorist who has spent two decades arguing that a sufficiently powerful AI would wipe out humanity.

The accelerationists are betting on speed

Opposing them are the accelerationists, who believe in the advancement of the technology, its inevitability and its benefit to society. They also make their fortunes from its success.

Marc Andreessen, who heads the venture capital firm Andreessen Horowitz, and David Sacks, a former White House AI czar, are the most prominent figures. Both are aligned personally and philosophically with technology investor Peter Thiel. The group opposes regulation and fights publicly against what it calls a well-financed doomsday cult.

Their argument is that safety-driven regulation is a front for companies trying to use the law to crush smaller rivals. Garry Tan, who runs the startup incubator Y Combinator, has made that case repeatedly, saying smaller companies cannot absorb the cost of compliance.

Andreessen and his allies fund political action committees that donate to Republican and Democratic lawmakers who promote AI growth. Anthropic is now trying to match that spending through its own fundraising.

Jensen Huang, the chief executive of Nvidia, sits in the same camp. His company’s chips are the bricks and mortar of the AI revolution, housed in the large data centres now meeting opposition across the United States.

Yann LeCun, who shared the Turing Award with Hinton and Bengio, has spent years dismissing their warnings of existential danger as overblown.

Big tech is walking a line

Most large technology companies try to proceed carefully. That is a positioning typical of corporations that do not want to be caught on the wrong side of a political fight.

OpenAI was founded in 2015 with a structure meant to ensure researchers, not investors, had the final say on how the technology was deployed, with safety for humanity as the guiding philosophy. Under Altman, who survived an internal challenge in 2023 led by safety-focused colleagues, the company now speaks to AI’s dangers while still accelerating.

Google, Microsoft and Amazon have long experience of talking about safety in technology while navigating Washington to keep regulation to a minimum. With AI, they are still walking that line.

What this means in South Africa

The argument matters here because South African companies are already buying the technology and the chips that run it. Banks, insurers, retailers and telecommunications operators are deploying AI systems for credit scoring, fraud detection and customer service. Each one is making the same choice the global players are making, between speed and caution, usually without a public debate about it.

The country has its own regulatory question. The Department of Communications and Digital Technologies has been working on an AI policy framework, and the Competition Commission has examined how digital platforms use algorithms. Neither has yet produced binding rules equivalent to the European Union’s AI Act.

South African firms that build AI products for export markets will face those foreign rules regardless of what Pretoria decides. That gives local companies a practical reason to follow the safety argument, beyond the philosophical one.

The cost of the technology is also a factor. Nvidia’s chips are expensive and the data centres that house them consume large amounts of electricity, a constraint South Africa knows well.

The debate is not settled

The three camps disagree about almost everything except the importance of the technology. The safety researchers say the risks are existential. The accelerationists say the risks are overstated and the benefits are being slowed. The large corporations say they are managing both.

For a South African reader, the practical question is simpler. The technology is arriving, the rules are still being written, and the companies selling it have not agreed among themselves on what those rules should say.

Source: eNCA, Doomers and boosters: who’s who in the AI wars, 15 September 2026.