Make a wish: OpenAI publishes blueprint for regulation

America should maximize the benefits of AI, writes OpenAI in a draft on regulatory issues. The technology is leading to reindustrialization.

Save to Pocket listen Print view
The OpenAI logo on the facade of the office building in San Francisco.

(Image: Shutterstock/ioda)

5 min. read

OpenAI compares artificial intelligence with the invention of the car. Apparently, America did everything right back then – compared to Europe, says OpenAI. They want to repeat that. The AI startup believes that developers who advance AI would pull the entire economy along with them so that everyone would benefit. The blueprint, which is aimed at the government, also states that this is an opportunity for reindustrialization.

Cars were invented in Europe. All experts are likely to agree on this. However, OpenAI believes that Europe has taken the wrong path. The speed limit and the Red Flag Act were introduced – but only in the UK. This law stipulated that someone had to run in front of a car with a red flag to warn other road users.

“America, on the other hand, took a very different approach to the car, bringing together private sector vision and innovation with that of the public interest to unlock the technology and its economic benefits – even in the face of the threat of the First World War.” This led to the USA becoming the “heart of the global automotive industry”. The German automotive industry might see things differently.

And the other conclusions drawn by OpenAI are probably not acceptable to the majority. In the USA, unlike in Europe, the government has supported the switch from horse-drawn carriages to cars instead of giving priority to carriages. In addition, investments have been made in better roads – at federal level and not with individual state road networks. Although there is currently a decree issued by US President Joe Biden, who is still in office, the states can enact further laws. There are now said to be more than 700 of these. Of course, there is also the analogy to infrastructure and the question of energy, which AI requires.

OpenAI CEO Sam Altman is also not a supporter of laws such as the Chips Act. Although this is actually intended to benefit the US industry, according to Altman it has not led to the desired successes. Biden recently presented the so-called “Export Control Framework for Artificial Intelligence Diffusion”. This is also intended to protect the US economy and security interests through export restrictions. For example, chip designers are only allowed to sell their advanced GPU accelerators to 18 countries without restrictions – including Germany, but not China.

In the blueprint, Chris Lehane, Vice-President Global Affairs at OpenAI, also warns that if the USA does not win the race for supremacy in the field of AI, China will win and the Chinese government will gain global influence. Maintaining the paragraph that Altman has an appointment in Washington D.C. on January 30, i.e., after Donald Trump takes office as the next US president, the draft states: “We believe in America because America believes in innovation.” Chips, data, energy, and talent are the keys to victory. America will win the race.

To summarize, the draft specifically calls for AI to be subject to little regulation, even if only uniformly at federal level. Although OpenAI emphasizes that AI will benefit everyone, it then writes that the blueprint should only deal with large language models. AI applications, such as those used to detect cancer at an early stage, are not included. In addition, the US government should ensure that AI companies can use a secure infrastructure, such as classified computer clusters, to assess security risks and protective measures. How it is to ensure this remains open.

It gets more specific here: The US government should make it easier for authorities to access AI services so that taxpayers can benefit from a leaner structure. Donald Trump is also planning to cut jobs within the government. So instead of paying people for their work, the state should rather pay OpenAI to provide software.

On the other hand, it seems sensible to prohibit AI companies from offering services that make it possible to create content that shows child sexual abuse (CSAM). OpenAI also considers mandatory watermarking to be desirable. Finally, there are several calls to train people and develop a national AI strategy for education.

As a blueprint, the paper is a “living document”, so it can be further elaborated and amended at any time.

(emw)

Don't miss any news – follow us on Facebook, LinkedIn or Mastodon.

This article was originally published in German. It was translated with technical assistance and editorially reviewed before publication.