The proliferation of artificial intelligence (AI) applications in recent years has highlighted the importance of regulatory frameworks to ensure AI’s responsible use. Recognizing this, the EU has become the first global power to pass AI-related legislation. The EU AI Act, considered a watershed moment in today’s digital era, is expected to create ripples worldwide and prompt leaders to take initiatives to control the use of AI.

The AI industry, valued at US$454.1 billion in 2022, is predicted to reach US$2.6 trillion by 2032, according to a 2023 report by Canada/India-based market research company Precedence Research. Although this impressive increase in the use of AI offers immense potential, it has raised numerous concerns about its misuse. Many industry experts have voiced concern about how significantly AI impacts important industries such as education and health, and may eventually alter human lives.

Regulatory bodies and governments worldwide are also now aware of the risks of bias, discrimination, and privacy breaches that come with the unrestricted use of AI. A 2020 study published in Sustainable Development, an interdisciplinary journal, found that unchecked AI poses a threat to the Sustainable Development Goals (SDGs) established by the UN.

The EU took its first step in addressing concerns related to AI in April 2021 when it released the first draft of EU AI regulation. However, this draft was revised after the 2022 release of ChatGPT to include the newer technological advances with a future-proof approach that will enable the law to evolve as technology does.

The EU AI rule uses a risk-based strategy to divide AI systems into categories, namely unacceptable, high, limited, and minimal risk. Systems categorized in the unacceptable risk group will be banned, and those with high risk will undergo a rigorous assessment to understand their effect on fundamental rights and be given a CE mark before their market release.

The limited risk category that fulfills specific transparency requirements should follow EU copyright laws, prepare technical documentation, and release a synopsis of the training materials so the users can make an informed decision. Companies can release minimal-risk systems, such as spam filters and AI-powered video games, without restrictions.

The AI Act has also introduced certain transparency requirements for general-purpose AI (GPAI) models such as Gemini and ChatGPT. For powerful and highly effective models, there are additional risk management requirements, such as maintaining cybersecurity standards, conducting evaluations, assessing and mitigating risks, and reporting serious events.

The EU AI Act has several implications for business across the continent

Many industry experts consider the EU AI Act a significant regulatory tool for overseeing the advancement and utilization of AI technologies throughout the continent. The enforcement of this act is expected to influence significantly the operations, approaches, and competitive environment across several sectors in the EU, as well as intercontinental business with products traded in the European market.

Achieving compliance is one of the main challenges businesses will face with the introduction of the new EU AI law. The law comes with a hefty penalty for non-compliance, with most violations costing businesses €15 million, or 3% of their annual global turnover. Non-compliance concerning banned AI systems can result in fines of up to €35 million, or 7% of the company’s annual turnover. Furthermore, providing false, deceptive, or incomplete information may result in fines of up to €7.5 million, or 1.5% of the total turnover.

Any organization aiming for compliance needs to perform a gap analysis to identify disparities and enhance company processes, policies, and metrics. They must also provide the regulators with accurate, efficient, and timely answers. All these place a substantial organizational and economic burden on the companies.

Another challenge businesses can face is implementing all the changes needed to fill the compliance gap while being consistent with their internal structure. This will require the company to identify the metrics it needs to track to achieve compliance and design new organizational procedures to fulfill this. This should also be done in such a way that it does not hinder other strategic goals, such as innovation, budgetary constraints, etc., placing additional strain on the leadership.

Companies offering multiple AI solutions can face several complicated roadblocks with the implementation of the EU AI Act. Such organizations will be subject to a different set of regulations depending on the risk category of their AI products. This can lead to internal policy confusion.

Slower product development cycles and delays in product releases are other bottlenecks that companies will need to address with the act’s implementation. New AI-powered products, especially high-risk ones, now need to undergo more rigorous evaluation to ensure compliance, which can slow the entire process.

Players can also face challenges in M&A activities with the new regulations. Businesses will now need to spend more time and resources assessing the compliance of their merging partner before proceeding.

There are several opportunities for determined businesses

While the implementation of the EU AI Act does spell several challenges for businesses, it also offers opportunities for interested players. With the new act in place, customers will be able to place more trust in AI solutions. This will enhance the adoption of new AI-based systems.

Determined players can also improve innovation and gain investment with the help of Article 53 of the Act. This section states the possibility of establishing “regulatory sandboxes” that can be set up by one or more member states. These sandboxes offer controlled environments for developing, testing, and validating new AI technologies for a short time under the guidance of a competent authority. This will also ensure that the AI solutions fulfill regulatory compliance.

The EU AI Act offers special support measures to start-ups and SMEs. The compliance requirements for high-risk AI have been modified to make them less burdensome and technically feasible. Start-ups and SMEs also get a proportional cap on compliance infringement fines. This will make it easier for budding businesses in the AI sector to make leaps in innovation and product development.

Interested EU-based players can also receive support for product development through programs such as Testing and Experimentation Facilities, Data and Robotics Hubs, Digital Innovation Hubs, etc. The AI Office, set up by the EU AI Act to oversee the rules and regulations related to GPAI models, has established forums where stakeholders can exchange best practices and contribute to rules of conduct and practice. This can improve participation across industries and enhance inclusiveness.

EOS Perspective

The EU AI Act can be considered a landmark development in the regulation of AI technologies in Europe. It has extensive implications for businesses, society, and the economy on the continent and worldwide. Many industry leaders consider this act a starting point in AI regulation. Other countries are expected to follow in the EU’s footsteps soon and make similar laws curbing the effects of unregulated AI.

The EU AI Act is expected to make AI-based solutions safe and bias-free, with better transparency into their developmental processes. It is also expected to enhance accountability and create a more responsible approach to AI development and deployment.

Businesses, especially SMEs and start-ups, can expect several benefits from this act. Experts predict that with the renewed focus on safety and ethical issues, there will be large-scale development of far more trustworthy and robust AI-based solutions in the future.