2024's Key Legislation on AI in Radiology: What You Need to Know

Authors: Ovidiu Coruga , Bianca Maria Dascalu

Governments all around the world are developing and putting into effect governance laws and regulations in proportion trying to keep the speed of AI-powered technology that is emerging. Finding a balance between innovation and regulation is a challenge for governments, given the revolutionary nature of AI technology.

As the integration of Artificial Intelligence accelerates, understanding the evolving legislative landscape becomes essential for healthcare providers. Let’s dive into recent legislation in key regions over the globe and the implications for the future.

United States: A Framework for Risk-Based Regulation

In the United States, the approach to AI regulation is increasingly adopting a risk-based framework. This framework, proposed by the National Institute of Standards and Technology (NIST), categorizes AI applications based on the level of risk they pose, with each sector and agency tailoring implementation to their specific needs. The outcome of the 2024 presidential election is poised to significantly influence the regulatory discourse, especially in terms of managing AI's impact on information dissemination and privacy concerns.

The Office of Management and Budget (OMB) has proposed guidance that leverages the AI RMF and the Blueprint for an AI Bill of Rights. This guidance mandates evaluation, monitoring, and risk mitigation practices tailored to the context of the federal government. It highlights the strengthening of AI governance, advancing responsible AI innovation, and managing risks from using AI, by requiring federal agencies to designate Chief AI Officers and establish AI Governance Boards.

International cooperation and alignment are also significant concerns for the U.S., as it aims to set a global standard for AI regulation. Efforts are being made through various international forums and partnerships, such as the US-EU Trade and Technology Council and the Global Partnership in AI, to ensure that the U.S. can effectively lead on international AI governance issues.

Federal agencies with regulatory purview, like the Food and Drug Administration (FDA), have already begun incorporating AI, having cleared over 700 AI-driven solutions, particularly machine learning, into medical devices, publishing best practices, and advancing regulatory science through an AI action plan.

European Union: Pioneering Comprehensive AI Legislation

The EU has made a landmark move with the AI Act currently, positioning itself as the first major jurisdiction to enact sweeping legislation governing AI. This act introduces a risk-based categorization of AI systems into four tiers, focusing on the sensitivity of the data involved and the specific use cases or applications of AI.

The AI Act mandates increased transparency and accountability for developers and users of high-risk AI systems. It sets forth requirements for documentation and auditing to minimize biases and ensure the representativeness of training datasets. The AI Liability Directive, still under negotiation, aims to establish mechanisms for financial compensation for harm caused by AI, further emphasizing consumer protection and accountability.

The legislative process has seen areas of potential disagreement among EU bodies, particularly concerning the definition of high-risk AI, governance structures, and the regulation of AI for biometric surveillance. The collaborative efforts aim to resolve these differences and establish a governance mechanism that includes a new European AI Office to coordinate compliance, implementation, and enforcement, with advisory support from a scientific panel of independent experts.

Penalties for non-compliance are significant, ranging from €7.5 million or 1.5% of global turnover to €35 million or 7%. The act's enforcement mechanism underscores the EU's commitment to establishing a rigorous regulatory environment for AI.

China: Moving Towards Comprehensive AI Regulation

China's approach to AI regulation is shifting towards a more unified framework with the announcement of an artificial intelligence law by the State Council. This move signifies a departure from the previously fragmented regulatory environment, where rules targeting new AI applications were developed in an ad-hoc manner.

The proposed national AI law is expected to cover all aspects of AI, from development to deployment. A significant aspect of this new framework is the establishment of a National AI Office, tasked with coordinating and supervising the administration of AI technology across the country. This office is designed to streamline and centralize AI regulation, preventing the overlap of authority among various regulatory bodies.

Part of the law's objective is to implement a "negative list" system for high-risk AI research, including areas like radiology and healthcare, which would require government approval before proceeding. This approach is intended to balance the promotion of AI innovation with the need to address potential risks associated with certain types of AI applications.

The interim measures introduced by the Cyberspace Administration of China (CAC) in July 2023, which precede the comprehensive AI law, underscore China's intention to encourage AI innovation while retaining control over the information disseminated by AI technologies. These measures outline key provisions for generative AI, emphasizing the need for AI products and services to align with core socialist values and undergo security reviews before public release.

Japan`s AI regulations influence radiology software development

The way that Japan approaches regulating AI could have a big impact on how world leaders come to agreements. Rather than stifling AI due to exaggerated concerns, it has created and updated laws on the technology to optimize its beneficial effects on society. Decisions are made by assessing the potential risks and including input from different stakeholders, rather than relying on a requirement for everyone.

In general, Japan has no laws limiting the application of AI. According to the AI Governance in Japan Ver. 1.1 report published by the Ministry of Economy, Trade and Industry (METI), the requirements for AI systems are not necessary for the moment. This is a result of the inability to keep up with the complexity and speed of AI evolution. In this situation, a rigid, thorough, and prescriptive policy could hinder innovation. The METI report concludes that while offering non-binding guidelines to support or steer such initiatives, the government should respect businesses' voluntary efforts for AI governance.

The primary legal requirements for organizations that gather, utilize, or transfer personal information are outlined in the Act on the Protection of Personal Information (APPI). In 2022, the most recent update to the APPI established the notion of pseudonymized personal data. This new idea is anticipated to encourage firms to use more data for AI research because the duties for handling pseudonymized information are less onerous than those for handling personal information.

United Kingdom joins the battle with new AI Regulations

For the first time, the US and the UK have agreed to work together on testing and evaluating the dangers associated with new AI models, making them the first nations to do so.

The deal, which US Commerce Secretary Gina Raimondo and UK Science Minister Michelle Donelan signed on Monday, April 1st, in Washington, outlines how the two countries would collaborate to share technical expertise, and data on AI safety. The agreement enables the UK's newly formed AI Safety Institute (AISI) and its yet-to-begin counterpart in the US to exchange knowledge by temporarily assigning researchers from both countries. These institutes will work together to create objective evaluation methods for private AI models made by companies like Google and OpenAI.

Researchers including Chris Summerfield from the University of Oxford and Geoffrey Irving from Google DeepMind have been employed by the UK government-backed AISI, which is chaired by tech investor and entrepreneur Ian Hogarth, to begin testing both pre-release and current AI models. Among the tech companies that voluntarily committed to releasing their most recent generative AI models for assessment by Britain's AISI, which was founded in the wake of the UK's AI Safety Summit at Bletchley Park, were OpenAI, Google DeepMind, Microsoft, and Meta.

Germany proposing AI Regulations since 2018

The National AI plan, made by The Federal Ministries of Education and Research, Economic Affairs and Energy, and Labor and Social Affairs, was introduced by the German Federal Government in November 2018 since Germany saw the potential of AI evolution. The strategy's array of policy efforts seeks to accomplish the following objectives: enhancing and strengthening Germany's competitiveness going forward by positioning Europe and Germany as leaders in AI; ensuring responsible AI development and application that benefits society; integrating AI into society in terms of ethics, law, culture, and institutions within the framework of an extensive social discourse and proactive political actions.

The Study Commission on Artificial Intelligence was a legislatively constituted group that regularly looked into AI-related regulatory matters. The Study Commission makes actionable recommendations in its final report. The Study Commission advocates for sector-specific AI regulatory frameworks that uphold the proportionality and liability criteria. The German Federal Government started steps to address challenges connected to information management, data ownership, free flow of data, and standardization.

Poland, Turkey, and Greece join AI regulations

Like many other EU member states, Poland has not yet passed any legislation governing artificial intelligence. This is because lawmakers are awaiting the adoption of pertinent EU regulations, such as the Artificial Intelligence Liability Directive and the AI Act, which we have already covered previously.

The Ministry of Digital Governance (MDG) is leading the development of Greece's national AI strategy, collaborating with key stakeholders from the EU and Greece, including experts in healthcare AI and radiology. To ensure the safe development and use of AI, the group is trying to construct a strong framework that includes data policies, ethical principles, and a balance of skills and trust.

In Turkey, the strategy is made by the Digital Transformation Office of the Presidency of the Republic of Turkey (DTO) for 2021-2025. The Strategy covers the principles of AI, with a particular focus on the establishment of human-centered AI. According to this argument, AI should support the preservation of democracy, human rights, and the rule of law in addition to improving the environment, encouraging diversity and inclusion, and creating peaceful communities.? AI values and principles must be followed while handling personal data used for AI systems at every stage of its lifespan, including collection, processing, transfer, storage, and deletion.

Global Trends and Implications for Radiology

In the spirited journey through the world of AI and its implications over the realms of radiology and healthcare, different nations are paving the way with their legislative frameworks. Now it's clear that as AI technologies evolve, so too must our approaches to governance and regulation to ensure that innovation can flourish while maintaining safety and ethical integrity.

As we appreciate the incredible potential AI holds for transforming healthcare—making diagnostic processes more precise, treatments more personalized, and patient outcomes better than ever before, the true success hinges not just on technological advancements, but on our ability to adapt and create supportive environments through thoughtful legislation.

What are the next steps? As we stand on the brink of this revolutionary shift, healthcare providers, legislators, and innovators need to come together more frequently and in greater numbers. Understanding the legal landscapes and preparing for the changes they bring will ensure that we harness AI's full potential responsibly and effectively.

Do you want to delve deeper into how AI is shaping the future of radiology and healthcare?

?????? Download your free Ebook Today: https://mindfullytech-8819877.hs-sites.com/choosing-the-best-ai-radiology-vendors