CHINA
19 September 2026
The US must win the AI race
David Sacks, Co-chair of the US President's Council of Advisors on Science and Technology, argues that the US must win the AI race.
This article is a resume of a recent CBS news interview with David Sacks, Co-chair of the US President's Council of Advisors on Science and Technology.
Many people – tech bosses, politicians, academics and civil society – are now panicking about the impact of AI. Sacks does not subscribe to the view that AI will end humanity, as some doomsayers claim . We are in the midst of a panic.
We're in an election season where there is an incentive for political actors to amplify the fear. We need to calm down and look at the issue rationally. AI safety is an issue, but it's something we can address and it's something that the AI labs themselves have to get right.
In sum, there is lots of noise about AI. But this technology exists. If the US stops developing AI, it's a certainty that China will develop, lead and dominate.
Company leaders have a responsibility to make sure their products are safe. Scaring the public, as they are doing, is not good. Leaders of frontier labs have a responsibility to tell us how they are going to develop AI safely. We’ve reached the stage in AI development where safety (predictability, reliability, robustness) must be prioritised.
Anthropic CEO Dario Amodei is saying we need to pace the frontier. But Anthropic and OpenAI are the frontier, they are a duopoly. They should just do it, no-one is holding them back. AI lab CEOs seem to be trying to deflect responsibility.
In sum, we need Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman to take responsibility right now.
For example, Dario says that he needs antitrust waivers in order to pursue the AI safety agenda. Sacks thinks that Dario wants to form a cartel or be protected from competition. Some AI companies are trying to get a waiver on product liability.
For Sacks, that's not acceptable. He notes that there are millions of companies across America who know they have to make their products safe and they don't get protection from the government for doing that. You have an obligation to make your product safe. If you put out a product that's unsafe, you can be subject to civil lawsuits or even be subject potentially to criminal liability.
The AI leaders should be responsible, and \act responsibly. That is their obligation to the public. Any additional laws should be sensible, grounded in facts and evidence. People calling for more regulation should identify what is not already covered.
The US already has a lot of laws that pertain to AI. We live in a highly regulated administrative state and already have lots of statutes and regulations and regulatory agencies. They do everything from regulating harms like cyber attacks or fraud that could be perpetrated through AI, or child safety or privacy violations.
Sacks believes the US should talk to China about AI, but there is no hope of agreeing on a deal. The priority should be for Anthropic and OpenAI to do a good job in managing AI safety in their labs – they are the frontier. If they do this, a deal with China or other countries becomes a little bit less important.
Chinese President Xi Jinping will soon be meeting with President Trump. What is your advice to President Trump? Have a conversation about what they are thinking about AI. But Sacks has little faith that we're going to be able to negotiate something like that anytime soon.
Sacks does not believe in the argument for a transnational organisation to regulate AI. It would take years to set up, and we don't have years. Sacks believes the American government should have jurisdiction and control over AI in America. It should talk to other governments to see how they're thinking.
This is such a profound technology, for the economy, the military and dual use (military/civil). The US is in the lead now, and will stay in the lead as long as we just pursue continued development in a sensible way. If the US stops AI development as Bernie Sanders proposes, China will rocket past the US. If the US loses the AI race and China is the main competitor, then the US will no longer be the number one country in the world.
In response to the current AI hysteria, Sacks notes that a lot of recent dire predictions have not come true. A year ago we were being told that AI would lead to a jobs apocalypse. This has not happened. These doomsayers should be held accountable, and this should be kept in mind when they make their next prediction.
Further, the problems that we have experienced with AI are all things that should have been managed and can be fixed. If the companies don’t think they can fix these problems, that they don’t think they can make their models safe, they are not the right people to be working on these models and should step aside.
Many people – tech bosses, politicians, academics and civil society – are now panicking about the impact of AI. Sacks does not subscribe to the view that AI will end humanity, as some doomsayers claim . We are in the midst of a panic.
We're in an election season where there is an incentive for political actors to amplify the fear. We need to calm down and look at the issue rationally. AI safety is an issue, but it's something we can address and it's something that the AI labs themselves have to get right.
In sum, there is lots of noise about AI. But this technology exists. If the US stops developing AI, it's a certainty that China will develop, lead and dominate.
Taking responsibility for AI risks
Sacks sees a contradiction when leaders of frontier AI companies worry about humanity, while at the same time they continue to develop AI models. If you can’t control your AI models, don’t build them or step aside!Company leaders have a responsibility to make sure their products are safe. Scaring the public, as they are doing, is not good. Leaders of frontier labs have a responsibility to tell us how they are going to develop AI safely. We’ve reached the stage in AI development where safety (predictability, reliability, robustness) must be prioritised.
Anthropic CEO Dario Amodei is saying we need to pace the frontier. But Anthropic and OpenAI are the frontier, they are a duopoly. They should just do it, no-one is holding them back. AI lab CEOs seem to be trying to deflect responsibility.
In sum, we need Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman to take responsibility right now.
The risk of regulatory capture
Sacks believes that when tech leaders come together and argue for regulation, it can be self-serving. It is potentially a way for them to make it harder for competitors to get into the race. There is the well-known phenomenon of regulatory capture. Companies go to the government to seek regulation and try to bend the regulations to their advantage.For example, Dario says that he needs antitrust waivers in order to pursue the AI safety agenda. Sacks thinks that Dario wants to form a cartel or be protected from competition. Some AI companies are trying to get a waiver on product liability.
For Sacks, that's not acceptable. He notes that there are millions of companies across America who know they have to make their products safe and they don't get protection from the government for doing that. You have an obligation to make your product safe. If you put out a product that's unsafe, you can be subject to civil lawsuits or even be subject potentially to criminal liability.
The AI leaders should be responsible, and \act responsibly. That is their obligation to the public. Any additional laws should be sensible, grounded in facts and evidence. People calling for more regulation should identify what is not already covered.
The US already has a lot of laws that pertain to AI. We live in a highly regulated administrative state and already have lots of statutes and regulations and regulatory agencies. They do everything from regulating harms like cyber attacks or fraud that could be perpetrated through AI, or child safety or privacy violations.
Competition with China
We should not pause or stop AI development because of the perceived risks of AI. That would hand the race to China. For chips, China is a couple years behind the US, for AI models about six to 12 months behind. That’s a decent-sized lead, but not if the US pauses AI development. If the US pauses AI development, China will race ahead. China is not going to stop AI development. They don't have these like doomer groups telling the population that they're going to go extinct in a few years.Sacks believes the US should talk to China about AI, but there is no hope of agreeing on a deal. The priority should be for Anthropic and OpenAI to do a good job in managing AI safety in their labs – they are the frontier. If they do this, a deal with China or other countries becomes a little bit less important.
Chinese President Xi Jinping will soon be meeting with President Trump. What is your advice to President Trump? Have a conversation about what they are thinking about AI. But Sacks has little faith that we're going to be able to negotiate something like that anytime soon.
Sacks does not believe in the argument for a transnational organisation to regulate AI. It would take years to set up, and we don't have years. Sacks believes the American government should have jurisdiction and control over AI in America. It should talk to other governments to see how they're thinking.
AI optimism
All things considered, Sacks is optimistic about AI. The United States is a country of innovators. It basically invented electricity and the electrical grid, computers, the personal computer and the Internet. Now the US has got this AI wave started. The US should try to stay in the lead.This is such a profound technology, for the economy, the military and dual use (military/civil). The US is in the lead now, and will stay in the lead as long as we just pursue continued development in a sensible way. If the US stops AI development as Bernie Sanders proposes, China will rocket past the US. If the US loses the AI race and China is the main competitor, then the US will no longer be the number one country in the world.
In response to the current AI hysteria, Sacks notes that a lot of recent dire predictions have not come true. A year ago we were being told that AI would lead to a jobs apocalypse. This has not happened. These doomsayers should be held accountable, and this should be kept in mind when they make their next prediction.
Further, the problems that we have experienced with AI are all things that should have been managed and can be fixed. If the companies don’t think they can fix these problems, that they don’t think they can make their models safe, they are not the right people to be working on these models and should step aside.