Navigating the European AI Act and Regulatory Frameworks
AI compliance challenges
Sean Musch on meeting EU AI Act regulations
The latest updates on AI legislation
Upcoming AI events and conferences
Before you go…check out the latest news about AI regulation
“With GDPR, the regulation wasn't the problem. The problem was categorizing and understanding your data. With the European AI act, it's going to be understanding and categorizing how…you use your AI systems.” — Sean Musch
The latest episode of the RegulatingAI Podcast was recorded live at the World Summit AI.
In this episode, we dive into the complexities of AI compliance and the challenges organizations face in navigating the evolving regulatory landscape, especially with the European AI Act. Learn how businesses can stay compliant while driving innovation in AI development.
Our guest, Sean Musch, Founder and CEO of AI & Partners, shares his expertise on the European AI Act and other regulatory frameworks shaping the future of AI. Discover practical strategies for navigating compliance while fostering responsible AI practices.
Tune in to hear this conversation, and keep reading for a summary!
In this episode:
Regarding what AI & Partners does: They found out that when they were trying to evaluate companies’ AI, they had the hardest time identifying all the models they were using. “For example…Google Translate has a little bit over 40 models in it. So that means basically every model needs to be individually assessed,” explained Musch. They can go into the cloud and identify all your algorithms, AI models, LLMs, and so on, and self-assess them according the necessary regulation.
Where do we stand in terms of a regulatory framework? “What we found out is actually every company wants to have their own framework. So there is no one fits all,” he said.
They also found that some companies want to have more industry diagnostic frameworks in place.
He also said about frameworks that a framework is just about following the regulation. We don’t need to complicate it much more than that.
What should companies that need to comply with the EU AI Act be thinking about? “You can actually expect the same flow as GDPR, in my opinion,” said Musch. Healthcare will need to be among the first to comply, because of the risk involved, and it will likely be a struggle for those companies.
A lot of companies need to understand that they do use AI. An organization could have thousands of AI systems, because from a regulation standpoint, things like chatbots and spam filters meet the legal definition.
“Basically, everything that does automated decision-making based on data is an AI system,” he said. “With GDPR, we tried to regulate the data, and you see the data as gasoline. And with the European AI Act, we're regulating the combustion engine.”
You have to be aware of the AI in your whole supply chain, too. For example, if you have a French voice recognition system, and you put that into a medical device made by an Indian company, that means the Indian company needs to be EU AI Act-compliant.
There’s a fragmentation of AI regulation. There’s the EU AI Act, but the U.S. doesn’t comprehensive federal AI regulation. How do we deal with that? For smaller companies and startups, it’s simple. You already know what your high-risk models are. And you know what you use and how you use it.
What will be the role of AI audits looking into the future? Will audits happen across companies? “Oh yeah, 100%,” said Musch.
“With auditing, of course, you have a lot of samples,” he said. “One thing that is required by the EU AI Act is every AI system needs to have a self assessment.” You can take the sample sizes and check that they’re categorized the right way.
Companies want to be sure as much of their AI as possible is below “high risk.” That’s because “everything under ‘high risk’ is going to be publicly available for the audit,” he said.
With the many companies you talk to, is there a sense of urgency to be in compliance? “It really depends also the industry,” said Musch. In hospitals, for example, there’s more urgency to be compliant.
“For finance, we see some companies—especially the fintechs—we see a bigger urgency,” he said. “But it is also still a lot of discovering, and still a lot of understanding for a lot of companies [about] how does it impact my business?”
How careful should companies and people be in terms of data transparency and bias? Checking for bias is a big focus of the EU AI Act.
A lot of what they do is look for exceptions. And a challenge is the sheer number of systems. “And it's just so much work to identify, understand how you use it, categorize it—and that's where we help,” said Musch.
“With GDPR, the regulation wasn't the problem. The problem was categorizing and understanding your data. With the European AI act, it's going to be understanding and categorizing how…you use your AI systems,” he said.
Do you think synthetic data will help protect from inherent biases? It depends a lot on the industry. Musch said because of that, he couldn’t give a good answer. But, “there are examples [where] it really works and there are examples [where] it really doesn't work,” he said.
What do you see as the role of regulations in the future, as these technologies evolve so rapidly? “The good thing is, regulation is for everybody,” said Musch. So if you’re in a given industry, everyone is that industry has to deal with the same regulations. So that evens out the playing field.
That’s why regulations like the EU AI Act define AI so broadly. “They cannot regulate as fast as the technology goes,” he said.
What actionable advice would you give listeners/readers? He suggests doing your company self-assessment, understand it (eg, your risk levels), and understand your legal challenges. Don’t look at your legal challenges first before you know what you actually have.
Hear the entirety of Sean Musch’s conversation with Sanjay Puri on Apple Podcasts and Spotify. Subscribe to the RegulatingAI podcast to never miss an episode.
The latest AI legislation
H.R.12081, Washington, D.C. US Leadership in AI Standards. Advances U.S. leadership in technical standards by instructing the National Institute of Standards and Technology and the Department of State to take specific steps to facilitate U.S. involvement in creating standards and specifications for AI and other emerging, critical technologies. Status: Introduced.
S.5539, Washington, D.C. AI Review. Requires systematic review of AI systems before deployment by the Federal Government. Status: Introduced.
S.B.7676, NewYork, USA. Contract Requirements. Establishes requirements for contracts involving digital replicas of individuals, ensuring clear intended-use descriptions and representation to protect against unauthorized or exploitative use. Status: Enacted.
Check out our live AI legislation tracker
Upcoming AI events
International Conference on Artificial Intelligence – (ICAI-25)
Bayside, New York | Jan 1-2, 2025 | Online/In-person
Join us for the highly anticipated International Conference on Artificial Intelligence (ICAI-25) in New York, USA, on 1st – 2nd January, 2025. Hosted by Research Leagues, this conference serves as a premier global platform for sharing the latest advancements and developments in these fields. Our goal is to provide a competitive forum for practitioners and researchers from both industry and academia to come together and discuss cutting-edge advances.
The main objective of this conference is to offer attendees the opportunity to share their experiences and ideas on a global scale. Our program will feature industry-led presentations, panel discussions, and keynote speeches from international experts.
Infrastructure & AI Summit
Brentfod, UK | Jan 14-15, 2024 | In-person
Join us at the inaugural AI Infrastructure & Architecture Summit, where AI engineering leaders and infrastructure experts come together to redefine how enterprises design, deploy, and scale AI-driven applications. As an attendee, you’ll gain firsthand insights from experts on delivering enterprise-scale generative AI ecosystems through a purpose-built, full-stack platform. Learn how to manage AI compute resources effectively, meet AI demands at any scale with infrastructure designed for custom workloads, and stay ahead of the curve by adapting to the evolution of foundational AI models. This is a summit which will enable your teams to stay abreast of enterprise AI deployments and operational excellence.
This summit offers invaluable insights into how organizations can architect scalable AI ecosystems, unlock the power of custom LLMs, and harness AI infrastructure to drive growth and efficiency in today’s competitive environment. Whether you’re focused on operational excellence, innovation, or enterprise-scale transformation, the AI Infrastructure & Architecture Summit is your gateway to deploying AI-powered applications for the years ahead.
Davos-2025 AI Week: World Economic Forum Annual Meeting
Davos, Switzerland | Dec 26-29, 2024 | Online/In-person
The Annual Meeting 2025 convenes global leaders to address key global and regional challenges. These include responding to geopolitical shocks, stimulating growth to improve living standards, and stewarding a just and inclusive energy transition.
Bookmark this page to keep track of key upcoming AI events
Before you go…
AI Watch: Global regulatory tracker - United States | White & Case
Trump taps Sriram Krishnan for AI advisor role amid strategic shift in tech policy | CIO
The states are taking the lead on AI | Cato Institute
California imposes guardrails on the use of AI by payors for utilization management determinations | National Law Review
U.S. AI policy expectations in the Trump administration, GOP congress, and the states | Inside Global Tech
Sundar Pichai on AI, regulation, and what’s next for Google | NYT
Bipartisan task force sets health AI oversight priorities for the next Congress | STAT
What to expect in 2025: AI legal tech and regulation (65 expert predictions) | National Law Review
Bookmark this page to stay up to date on the latest AI regulation and policy news
Connect with experts, mentors, investors, and organizations to foster collaboration, innovation, and the responsible development of AI!
Find us on Discord, LinkedIn, Instagram, X, Facebook, and YouTube