Presentation by Salih MANSUR, Independent Researcher
The U.S. State of AI Policies and Regulations
Presented at the LMGlobal.Org‘s panel on “Towards a Public Discussion for Shaping the Future of A.I. Regulations in the U.S. & Beyond“, which virtually presented in collaboration with Bons-Ai.Org on November 15, 2025.
Presentation (PDF)

[LMGlobal.Org] The title of my presentation is “The U.S. State of AI Policy.” When I say ‘state,” I don’t mean an individual U.S. state, but rather the current condition of U.S. policies on AI.
My name is Salih Mansur. I’m a researcher. I completed my degree in Instructional Technology and I am now completing another Master of Science in AI Systems.

Before I start, I would like to thank The Light Millennium and all the organizers. I would also like to extend my gratitude to all inventors and innovators working on AI.
Because of AI, I have at least doubled my productivity. I have improved my work and myself. So I want to thank all parties involved.

I also want to say clearly: I do not claim to be an expert. I am learning together with you. I have learned a great deal today from Justin, from Mr. Adam Bricker, from Mr. Charles, from the Ambassador, and from everybody here.

Federal level: U.S. AI policy landscape

At the federal level, Justin has already gone over the U.S. state of AI policy very clearly and beautifully, so I will just highlight a few point.
Overall, there is no single AI Act in the United States like the EU AI Ac
This is striking, because the U.S. is the cradle of much AI innovation and many of the leading AI initiatives.

However, thanks to federal law and the U.S. system, the President can issue Executive Orders that shape how AI can be used and governed.
Other federal agencies can also act within their own domains – for example:

The Federal Trade Commission (FTC)

The Department of Education, which I follow very closely
The U.S. Department of Education is currently accepting input and recommendations on AI (for example, through early-December consultations). Each federal agency can start to set and refine its own rules.
But overall, this federal-level picture is not unified.
State level: fast but fragmented

Although the country is called the United States, much of its strength comes from diversity. States can act differently, experiment, and sometimes inspire each other.
At the state level, Justin also gave an excellent overview, so I will just share a few highlights I encountered while researching:

Utah has an AI Learning Lab to test AI and explore safe uses.
California has developed models aimed at safe innovation, trying to support innovation while protecting people.
New York recently introduced a bill on rent algorithms to help ensure that rental prices are not unfairly pushed up by opaque algorithms.
Other states have passed or proposed rules about:
Deepfakes
Hiring audits
Fairness and transparency in hiring practices
So, states tend to act quickly, but there is little coordination among them. In time, however, they tend to learn from and inspire one another.

Comparison with the European Union (EU)

I also looked at the European Union, which I follow through newsletters and email subscriptions.
Most of you know the GDPR – the data protection rules that pop up as cookie banners when you visit websites and ask you to accept terms.
GDPR comes from the EU, and
It is not fully replicated in the U.S. – only California has something similar.
In general, the EU is more unified in its regulatory approach than the U.S.
EU rules apply across almost all aspects of life.
Rules often apply to EU citizens even if they are not physically in the EU.
To sum up this part: The EU is more united and centralized in its AI and data rules. The U.S. is more diverse and fragmented, which can be a strength but also a challenge.

Canada: AIDA and human oversight

I was also interested in Canada, our northern neighbor. Canada has proposed the Artificial Intelligence and Data Act (AIDA) to help make AI safer.
One feature that I particularly liked is this:
If the government uses AI to make decisions, a real human must review the decision and make sure it is fair.
Recently, there was a case in the United States where a health department document was largely written by AI. It contained many broken links and non-working references. It seemed like an example where the “human-in-the-loop” check was missing.
In the Canadian approach, a system like that must go through human oversight before being accepted. I think this is a valuable principle.

Türkiye: emerging national AI strategy

I also looked at Türkiye, my homeland.
Türkiye started strongly in 2021 with a national AI strategy led by the Ministry of Technology.
They have created good regulations and strategic documents.

Progress is slow, but it is moving forward.
I was able to join some of their AI-related initiatives and trainings, and I even received a certificate for participating in one of their programs.
So Türkiye is building its own path, learning from global experiences while adapting to local realities.
Global perspective: the role of the United Nations
On the global level, I learned a lot from the United Nations.
Across the UN documents and discussions, you see repeated emphasis on:
Safety
Transparency
Human oversight
These themes connect national efforts and show that AI governance is not only a national, but also a global challenge.

Closing message

Through collaboration and cooperation, we can learn from one another. We do not need competition alone; we need coordination and shared learning.
In the end, we can all agree on a simple goal: Let us keep AI safe, fair, and human-focused.

Thank you.

_ . _

Presentation (PDF)