EU AI Act serves as a ‘guide’ and a ‘cautionary framework’ for businesses, says CEO Bernard Mallia

Equinox Advisory CEO Bernard Mallia has highlighted that the European Union’s (EU) Artificial Intelligence (AI) Act can serve as both a “guide” and a “cautionary framework” for businesses keen on introducing AI to their operations.

As part of its digital strategy, the EU has expressed a desire to regulate AI to ensure better conditions for its development and use. The aim is to make AI systems used in the EU “safe, transparent, traceable, non-discriminatory and environmentally friendly”, with AI systems having to be overseen by people rather than by automation, “to prevent harmful outcomes”.

Last month, technology ministers of nine Mediterranean countries pledged to support the adoption of the act during Techxpo in Ta’ Qali, stating that they supported the EU’s bid to promote the creation of “safe and lawful AI systems while promoting innovation in Europe”. These countries include Malta, Spain, Croatia, Cyprus, France, Greece, Italy, Portugal and Slovenia.

Following this, MaltaCEOs.mt reached out to Bernard Mallia, CEO of Equinox Group, to get his view about the EU’s AI Act as a business leader and IT specialist. Equinox Group is a Malta-based company that specialises in legal, corporate, re-domiciliation, deal structuring and compliance services, infrastructure and public utilities, as well as technology in terms of big data, business intelligence (BI), AI, and the Internet of Things (IoT).

He remarked that the EU’s AI Act represents a “significant milestone” in AI’s regulatory landscape, as the area “clearly needs to be regulated”, and this is a “step in that direction”. This is primarily because it addresses “critical issues” including transparency, accountability and data privacy.

“The act serves as a foundational structure that can guide both developers and users of AI in adhering to ethical standards and legal requirements. Its implementation would contribute to building public trust in AI systems, which is indispensable for widespread adoption and use,” Mr Mallia said.

However, he noted that it cannot be seen in isolation, as it needs to be analysed within the context of the General Data Protection Regulation (GDPR) and the Digital Markets Act (DMA), as each of these serves a specific purpose but has overlaps that “complement or complicate” each other’s objectives.

Despite this, he acknowledged that the act has some “fundamental flaws” that unless addressed, “could potentially undermine its effectiveness and the broader goals it aims to achieve”.

These include the potential of creating bureaucratic hurdles for SMEs, leading to increased compliance costs that “drive the general level of prices up unnecessarily” at a time when macroeconomic inflation is already at record highs. Compliance requirements are “onerous” for smaller businesses, and this can end up “stifling innovation” and creating an “undue competitive advantage” for larger businesses which have more resources.

Mr Mallia also highlighted the act’s lack of specificity, particularly the criteria in determining the risk levels of various AI applications. He said that while the levels are categorised, they are “not exhaustively defined”, and such vagueness, which is “possibly intentional” can lead to inconsistent interpretations and applications.

The act establishes obligations for providers and users depending on the level of risk from AI. This categorises them into minimal risk, limited risk, high risk, and unacceptable risk, with requirements increasing from one level to the next. Mr Mallia remarked that this is a “pragmatic” approach that allows for a nuanced regulatory framework that is adaptable. However, it is also crucial to ensure that these categories are “not static” and can “evolve with the technology”, something the AI Act has not yet addressed.

Mr Mallia also noted that the act “fails to sufficiently address” the issue of algorithmic bias, a “significant concern” in the deployment of AI systems.

“While it does touch upon data governance and quality, it falls short of providing robust mechanisms in relations to the audit and correction processes and obligations for biased algorithms. This is a critical gap, as biased AI systems can perpetuate existing social inequalities and create new forms of discrimination,” he said.

He affirmed that the act’s focus is “predominantly Eurocentric”, raising questions about its applicability and effectiveness in a global context.

“AI is a global technology, and its ethical and safety considerations often transcend national and continental boundaries. A more international perspective would strengthen the act and make it more universally applicable, without posing the risk of pushing those who are developing ethical AI technologies that are deemed high risk out of the EU to be able to continue developing them,” Mr Mallia added.

When asked what the EU’s AI Act means for businesses who are seeking to introduce AI to their operations, he said that it can serve as both a “guide” and a “cautionary framework”, as companies will need to conduct thorough risk assessments and adhere to compliance requirements. This is especially the case if the technologies being deployed are considered high risk.

“This could entail additional costs and time for implementation, but will also, it is hoped, result in more robust and trustworthy AI applications,” he remarked.

The act will also provide businesses with a legal clarity, reducing certain uncertainties which could hinder investment in AI technologies.

“Nevertheless, a certain legal of uncertainty will still remain due to the way the risk classifications have been categorised in operational area terms rather than on the basis of guiding principles,” Mr Mallia stressed.

During the meeting at Techxpo, the European Commission’s Director General for Communications Networks, Content and Technology Roberto Viola said that the Mediterranean and southern Europe tend to be “forgotten” when it comes to technological innovation.

Asked about this, Mr Mallia said that this observation is “as timely as it is accurate”.

“Addressing this disparity is a multifaceted challenge that requires a comprehensive and sustained policy approach with various organs of southern European governments working together for the same shared objective,” he said, before highlighting four key areas policymakers must tackle.

Firstly, he identified the creation of technology excellence clusters as an effective strategy. These would serve as “hubs for innovation”, bringing together academia, industry, and government to collaborate on “cutting-edge research and development”. Despite this, Mr Mallia said that the creation of such clusters is “not a trivial task”, as it necessitates a “well-thought-out framework of incentives” including tax breaks, grants and simplified regulatory procedures to attract start-ups and established tech companies to these clusters.

He also called for greater investment in education, particularly in science, technology, engineering, and mathematics (STEM) fields. This should not be limited to higher education but also primary and secondary levels “to build a strong foundation”.

“Specialised training programmes and workshops focused on emerging technologies like AI, blockchain, and cybersecurity could be introduced to equip the workforce with the skills needed for the future. Regrettably, education in southern European countries has not only lagged behind, but it has also deteriorated over time. This is not simply a matter of throwing money at the problem, but requires strategic and critical thinking to set up the right roadmap,” Mr Mallia emphasised.

Access to capital is another “critical factor”, as policymakers must work to create an environment that is “conducive to various forms of financing for tech start-ups”.

Lastly, he also pointed towards the need for a regulatory environment which is more “agile, manageable, and conducive to innovation”. “Policymakers should aim to reduce bureaucratic red tape that often hampers the rapid development and deployment of new technologies,” he said,  adding that this also includes consulting with industry stakeholders to “identify and eliminate any potential roadblocks”.

“Unfortunately, the EU’s AI Act as proposed fails to do that,” Mr Mallia concluded.