From the moment we unlock our phones, browse the latest news, engage in our daily tasks, shop online, or choose digital entertainment, we are constantly assisted by AI, which is reshaping every aspect of our lives. The AI landscape is vast and continuously expanding, encompassing everything from innovative chatbots and sophisticated image-generating AI to advanced mobile applications.
A Historic Event: the First Act to Regulate the AI Explosive Market
This year has marked a transformative phase for AI-powered technologies and tools, with GenAI captivating the global audience. The industry is thriving, with its market value anticipated to reach nearly two trillion U.S. dollars by 2030. Therefore, the adoption of suitable legislation to regulate the market was imminent. The European Union has introduced the trailblazing Artificial Intelligence Act, a critical milestone in AI regulation. This landmark legislation, unprecedented in its scope, addresses the complexities of the burgeoning AI market across various industries and lays down the foundation for ethical and secure AI development.
On 8th December 2023, following 38 hours of intense negotiations, the Council of the European Union and the European Parliament reached a Provisional Agreement on the Artificial Intelligence Act. This act is a cornerstone in our history, establishing the first comprehensive framework to regulate the use of artificial intelligence in various applications, from ChatGPT to biometric surveillance systems.
“Historic! The EU becomes the very first continent to set clear rules for the use of AI. The AI Act is much more than a rulebook - it's a launchpad for EU startups and researchers to lead the global AI race”, wrote European Commissioner Thierry Breton on his page on X network. The EU Commissioner, responsible for a suite of laws governing digital spaces and search engines, stated that 100 people negotiated for three days “to seal the deal”. His remarks highlight the EU's ambition to lead the global AI race, supported by clear rules and innovation platforms.
Pope Francis Calls for Worldwide Treaty on AI Regulation
Pope Francis has advocated for a binding international treaty to regulate AI, emphasising the need to preserve human values and cautioning against a potential "technological dictatorship."
I urge the global community of nations to work together in order to adopt a binding international treaty that regulates the development and use of artificial intelligence in its many forms. - Pope Francis
In his message for the Roman Catholic Church's World Day of Peace, the Pope stressed the importance of both national and international oversight in regulating AI, which has capabilities ranging from engaging in human-like conversations to writing computer code.
Key Aspects of the AI Act
The AI Act places the EU at the forefront of AI regulation, aiming to protect the public from potential risks posed by rapidly advancing technology. Let’s take a closer look at the main points of the legislation:
-
High-Risk Systems: AI systems identified as high-risk, due to their potential impact on health, safety, fundamental rights, the environment, democracy, elections, and the rule of law, must adhere to stringent requirements. These include undergoing fundamental rights impact assessments to gain access to the EU market. The classification aligns with the OECD’s approach, focusing on risk.
-
Transparency for Lower-Risk AI: These systems face minimal obligations, primarily around transparency, such as clear labelling indicating AI-generated content.
-
AI in Law Enforcement: The use of real-time remote biometric identification in public spaces by law enforcement is restricted to specific purposes, such as identifying victims of serious crimes or preventing imminent terrorist threats.
-
General Purpose AI (GPAI) and Foundation Models: These systems must comply with transparency requirements, including technical documentation and adherence to EU copyright laws. They also need to provide detailed summaries of the data used for training algorithms.
-
Prohibited AI Practices: The Act bans certain AI applications, including biometric categorisation based on sensitive characteristics, untargeted scraping of facial images for recognition databases, emotion recognition in workplaces and educational settings, social scoring, and AI that manipulates human behaviour or exploits vulnerabilities.
-
Sanctions for Violations: Fines are set based on the nature of the violation, with considerations for SMEs and startups. However, non-compliance with these regulations can lead to significant fines, ranging from 7.5 million euros to as high as 35 million euros, or a percentage of the global annual turnover of the company, depending on the severity of the infringement.
How does the AI Regulations impact business?
Challenges and Opportunities:
The legislation brings about both hurdles and prospects, setting a precedent for global AI governance. Startups face challenges in meeting transparency and compliance demands, potentially straining resources. Intellectual property concerns arise from the mandated transparency in training data. However, there are opportunities too. The Act fosters innovation through regulatory sandboxes and real-world testing, offering smaller companies certain derogations. This balanced approach could lead to safer, more ethical AI development, opening new avenues for growth and innovation in the AI sector.
Business Impact:
-
Impact on Smaller Startups: The Act could significantly hinder the progress of smaller AI startups. It's believed that many large-scale AI models will be categorised as "high risk", requiring these startups to comply with stringent obligations, especially around transparency.
-
Obligations for High-Risk AI Models: AI models deemed high-risk will have to meet various obligations, including providing technical documentation, information to users, and ensuring a high level of cybersecurity protection.
-
Challenges for General-Purpose AI Providers: The law requires providers of general-purpose AI to have detailed documentation available for regulators and to provide public summaries of their training data. This could be particularly challenging for smaller startups in terms of legal overhead and compliance costs.
-
Intellectual Property Concerns: Transparency requirements, especially regarding training data, raise concerns about intellectual property, as startups might have to share information that could be valuable to competitors.
-
Extra Costs for Compliance: Startups will face additional costs related to compliance activities, such as 'red teaming' to test models against various risks, potentially requiring additional staff hires.
-
Uncertainty and Ongoing Negotiations: There is still uncertainty regarding the final details of the regulation. The technical aspects are being discussed behind closed doors, which is crucial for startups to understand the full impact of the regulations on their business.
-
Investor Confidence Despite Risks: Despite the challenges posed by the AI Act, investors in startups like Mistral remain optimistic about their future prospects, viewing the regulatory risks as manageable.
By introducing clear guidelines and accountability measures, the Act could encourage more responsible AI development and use, potentially leading to greater public trust in AI technologies. However, the final details of the legislation are still being worked out, with implementation set for the near future.
Understanding and Fostering Trust
Trust in artificial intelligence emerges as a critical factor for its integration into society. According to BSI's 'Trust in AI' Poll, 56% of respondents across nine countries see AI as a boon to healthcare, demonstrating high enthusiasm for its role in medical diagnosis and remote health management. However, a significant 79% stress the importance of trust for AI's effective use in areas like carbon emission reduction. Workplace adoption varies globally, with 38% using AI daily, yet a knowledge gap persists, as a third of users of AI-driven services like curated playlists are unaware of AI's involvement. Over half would trust AI in their job with proper training, while 61% advocate for international guidelines for safe AI use. This highlights the need for education and transparency in AI to build public trust and optimise its future applications.
What’s Next?
With the final details of the AI Regulations Act still being refined, businesses are advised to begin preparing for compliance. The Act’s application, expected by the first half of 2026, signals a significant shift in the AI landscape.
At Sentium, we are closely following these developments, poised to navigate the challenges and seize the opportunities presented by this new era of AI regulation.