# Are we marching towards a state-controlled AI? > Just a few days ago, Leopold Aschenbrenner, a former member of OpenAI’s superalignment team, wrote a series of essays in which he observes various facets… - URL: https://datafund.io/blog/are-we-marching-towards-a-state-controlled-ai/ - Published: 2024-06-07 - Author: Datafund - Tags: privacy, ai, datafund, ethical-data, responsible-ai ![Source: Dall-E, Datafund](/assets/blog/are-we-marching-towards-a-state-controlled-ai/01.png) *Source: Dall-E, Datafund* Just a few days ago, Leopold Aschenbrenner, a former member of OpenAI’s superalignment team, wrote [a series of essays](https://situational-awareness.ai/) in which he observes various facets of AI. One of them highlighted an issue that Aschenbrenner sees as critical (at least in the context of the US): the United States’ lack of attention to the security of AI labs. Aschenbrenner writes that since AI is just a large file on a computer, all malicious actors have to do is steal it and render billions of investments almost a non-critical factor. [Dario Amodei, CEO of Anthropic, reinforces this threat](https://youtu.be/Nlkk3glap_U?si=-5qMhMrXUFqgl4Sz&t=2781): *“Could we resist if it was a state actor’s top priority to steal our model weights? No, they would succeed.”* This scenario of AI model theft could have far-reaching consequences, Aschenbrenner wrtites. It could potentially triggering a reckless rush towards Artificial Super Intelligence (ASI) devoid of necessary safety precautions. He therefore suggests that weights and algorithms be treated as critical infrastructure, similar to how states protect defence companies, such as Northrop Grumman. He contends that the US must take substantial measures to protect its AI industry to maintain a competitive edge. While the idea of government-controlled dissemination of critical AI research is a complex and contentious issue, it is nonetheless interesting to look at some of the recent events and see if they could point at such a shift. In this blog, we’ll explore whether there is any indication that countries could actually seal the weights and algorithms away from the public eye. ## The global recognition of AI’s importance ![Source: AI Index 2024](/assets/blog/are-we-marching-towards-a-state-controlled-ai/02.png) *Source: AI Index 2024* [](https://aiindex.stanford.edu/wp-content/uploads/2024/05/HAI_AI-Index-Report-2024.pdf) AI labs and their researchers have long been aware of the potent tool they are developing and have started to keep their models’ weights close to the chest as trade secrets. Countries, on the other hand, are just now in the process of increasingly acknowledging the strategic importance of AI through legislation: - The US has taken significant steps to advance its AI sector with [an executive order](https://www.whitehouse.gov/briefing-room/presidential-actions/2023/10/30/executive-order-on-the-safe-secure-and-trustworthy-development-and-use-of-artificial-intelligence/) aimed at a safe, secure, and trustworthy development and use of AI and bringing back chip production to the US with huge subsidies. All of this is running on enormous stream of private investments - China was also [quick to regulate generative AI](https://www.technologyreview.com/2024/01/17/1086704/china-ai-regulation-changes-2024/) and is currently preparing an overarching regulation of the sector. The country’s focus is on AI models’ safety and security, development of domestic AI technologies, and self-reliance in semiconductor production. - In the European Union, the proposed [Artificial Intelligence Act](https://www.europarl.europa.eu/doceo/document/TA-9-2024-0138_EN.html) classifies AI systems based on their risk levels and imposes stricter requirements on high-risk applications. This regulatory framework is complemented by substantial private and public investments in AI research and innovation. - Japan is focusing on ethical AI development and collaboration between the government, industry, and academia to secure its technological advancements. The country is enhancing its AI capabilities while ensuring safety and transparency. - The Middle East is also making strides in AI, with countries like the UAE investing heavily in AI technologies and infrastructure to drive economic growth and innovation. This includes partnerships with major tech companies and the establishment of AI research centres. These are just a few of the examples that highlight how countries around the globe are acutely aware of the impact the quickly evolving tech has and could have in the future. The rise in the number of legislative frameworks around the world is in step with the trends in the wider AI industry, which is seeing massive inflows of money into the startup scene and burgeoning numbers of new AI companies. ## Securing the domestic AI industry While there is a clear increase in AI-related legislation, indications of a full-blown state takeover of the AI sector are less apparent and largely indirect. We focus here only on the US and China as two global leaders whose actions largely affect what the rest of the world does. The most notable state-level actions to date include the US export ban on the most advanced chips to countries like China and Russia and the establishment of [NSA’s AI Security Center](https://breakingdefense.com/2023/09/nsa-stands-up-ai-security-center-as-focal-point-for-guiding-ai-use-by-government-defense-industry/). One of the tasks of the newly-established centre is to *“ensure that malicious foreign actors can’t steal America’s innovative AI capabilities”* to execute state-sponsored cyber attacks. These measures are accompanied by [additional legislative proposals](https://www.tftc.io/us-bill-tightens-ai-export-limits/) to grant the Commerce Department greater authority to regulate AI technology exports​​. As is often the case in China, their strategy focuses on [asserting control without hobbling development](https://thechinaproject.com/2023/07/21/chinas-new-regulations-on-generative-ai-sets-restrictions-but-leaves-room-for-innovation/). It aims to do this through the development of domestic platforms and a focus on open-source technologies within the country. By promoting self-reliance in critical technology sectors, the goal is to mitigate the impact of foreign restrictions and maintain control over its AI advancements​​. These measures indicate that at present, both powers seem primarily focused on advancing their capabilities, with the US also aiming to curb China’s rapid progress and computing capacities. But the tide could turn quickly in case of a major breakthrough. As Aschenbrenner noted in his essay: *“Sure, who cares if GPT-4 weights are stolen … But the AI labs are developing the algorithmic secrets — the key technical breakthroughs, the blueprints so to speak — for the AGI right now.”* And with the [widening gap](https://www.scmp.com/news/china/science/article/3254398/two-sessions-2024-chinas-top-advisory-body-told-ai-gap-us-widening) between the Chinese and US AI companies, the former might decide to increase efforts to steal trade secrets, which in turn could spur the US to actively start putting up walls around its AI sector. How that would play out, is probably a contentious legal issue. AI labs already have legal problems when trying to protect their intellectual property because [it evades the definitions of patents and copyright](https://academic.oup.com/jiplp/article/16/3/247/6143561#267224829). Among the first things we could see is legislative demands for additional and [more stringent measures in cybersecurity in AI labs](https://www.rand.org/pubs/research_reports/RRA2849-1.html). ## Conclusion As nations worldwide acknowledge AI’s strategic value, there is a concurrent trend towards onshoring the production of critical components, often driven by competitive dynamics between geopolitical rivals. The future of AI development may indeed be shaped by these efforts to control and protect technological advancements, and the potency of upcoming AI models will play a key role in this evolving landscape. We could be witnessing the early stages of a more fragmented and state-influenced global AI arena, where innovation, collaboration, and state influence coexist in a delicate balance. If you liked this content, keep following us for more on: ## [X](https://x.com/DataFundProject) ## [Linkedin](https://www.linkedin.com/company/datafund) ## or get in touch via: ## info@datafund.io --- [Are we marching towards a state-controlled AI?](https://medium.com/datafund/are-we-marching-towards-a-state-controlled-ai-3ed88e49823f) was originally published in [Datafund](https://medium.com/datafund) on Medium, where people are continuing the conversation by highlighting and responding to this story. --- *First published on [Medium](https://medium.com/datafund/are-we-marching-towards-a-state-controlled-ai-3ed88e49823f) on 2024-06-07.*