Deloitte Survey: Scepticism towards EU AI Act

The ratification of the EU AI Act has fundamentally changed the situation for European companies: The regulatory requirements are mandatory for all EU countries and in most cases must be implemented within the next two years. Neither European companies nor providers from outside the EU can expect a postponement of the requirements or immunity from penalties in the event of violations as soon as their AI applications are used in EU territory.

Hesitation in dealing with the EU AI Act is dangerous. Depending on the extent of AI use in a company, implementing the requirements will entail considerable effort - especially as many organisations don't even know exactly how much AI they use.

Dr. Till Contzen, Partner Deloitte Legal | Lead Digital Law

There is a threat of a situation comparable to the implementation of the GDPR, when companies were faced with a huge task in May 2018 as they had barely utilised the two-year transition period. Non-compliance could result in severe penalties.

On a scale of 1 to 5, how intensively is your company engaging with the "AI Act"?

Non-compliance could result in severe penalties

However, it is questionable whether comparable failures by companies to comply with the EU AI Act will be forgiven by the supervisory authorities in light of the rapid development of AI technologies. The threat of fines is considerable: violations of the EU AI Act are punishable by high fines, and the use of AI in these companies is also delayed, which can have a significant negative impact on business development.

The fact that around half of companies have not yet worked intensively on preparing for implementation reflects the fact that many companies in Germany do not yet have the topic of AI on their agenda. On the other hand, the EU AI Act has made its way into German boardrooms like hardly any other regulation to date. Particularly in highly regulated industries such as the financial sector or the healthcare sector, German companies are used to regulatory requirements being incorporated into compliance processes and systems and becoming an integral part of their framework conditions for innovation.

Dr. Sarah Becker, Partnerin Deloitte I Lead Digital & AI Ethics

The new law is accompanied by great uncertainty

In general, the Deloitte survey conducted by Civey revealed great scepticism towards the new regulation. The picture is particularly mixed when it comes to legal certainty:

Do you think the "AI Act" of the EU will lead to more legal certainty in dealing with Artificial Intelligence?

A similarly mixed picture emerges when it comes to the question of whether the AI Act will lead to more or less trust in artificial intelligence: Here, around a third (34.9%) believe it will have a positive effect, while almost as many (30.8%) cannot recognise this. On the contrary - almost half of those surveyed (47.4%) see the AI Act as more of an obstacle to the development and introduction of applications based on artificial intelligence in companies, while only a quarter (24.1%) are convinced that the new regulations will help.

The consequences of this scepticism are clearly measurable:

How well is your company, in your opinion, prepared to implement the requirements of the "AI Act" of the EU?

Accordingly, 52.3 percent fear that regulation will restrict their innovation opportunities in the field of AI; only 18.5 percent believe that the regulation will have a positive impact on innovation opportunities.

Legal framework essential for trust in AI

The regulations of the AI Act, some of which are probably deliberately unclear, do not always make interpretation easy. However, this is a step forward compared to the difficulties of an unregulated state in which the limits can only be guessed and companies and the industry are left to their own fate.

Dr. Till Contzen

There are high expectations among the population regarding the benefits of using AI, but at the same time there is also great concern about the negative effects:

Appropriate regulation can create trust, but at the same time there is a risk of weakening innovative strength. The limits of regulation remain to be seen due to the many rules that need to be interpreted. In principle, however, regulation has a confidence-building effect simply by setting a framework. In this respect, there is no way around considering the ethical and legal consequences of AI innovation.

Dr. Sarah Becker

Further Deloitte study confirms the need for regulations

The new edition of the global AI study ‘State of GenAI in the Enterprise’, for which Deloitte surveyed almost 2,800 managers from 14 countries, including 150 managers from German companies, comes to similar conclusions at a global level. The respondents cited compliance with regulations, risk management and the lack of a governance model as the biggest obstacles to the introduction of AI.

Deloitte's global report shows that the adoption of Generative AI (GenAI) has reached a critical stage where rising expectations of transformative impact and scaling are proving challenging. Here, three quarters of respondents said their organisations have increased investment in their data estate to address issues such as data security and quality. At the same time, only 23 percent on average worldwide feel well prepared in terms of risks, governance and regulatory issues.

Another important parameter for AI development in companies is the largely neglected measurement of success: now that promising GenAI pilot projects have led to more investment and rising expectations, management and supervisory boards are beginning to look for returns on their investments. However, scaling AI projects quickly is proving more complex than expected and revealing new challenges. However, demonstrating the value of GenAI implementations is critical to further investment - and this is where many organisations struggle to measure the exact impact of implemented measures, according to the latest edition (Q3-2024) of the global report ‘State of Gen AI in the Enterprise’.

Published: September 2024