On 1 August, the Artificial Intelligence (AI) Act entered into force across all 27 EU Member States, becoming the world’s first comprehensive law to regulate artificial intelligence. In the wake of the AI Act, a so-called ‘AI Office’ has been established with the aim of implementing the Act – especially its provisions on general-purpose AI (GPAI) – fostering the development and use of trustworthy AI, and international cooperation. In this context, what can be expected from the AI Act’s entry into force? Is this European AI Office good news for startups and scale-ups? For startups, the AI Office appears as one of their last chances to make themselves heard.

Brace yourselves, here comes the AI Act
The long-awaited AI Act is set to harmonise AI standards for safety, ethics and transparency. Over the next 18 months, the Act will gradually implement specific rules: provisions on prohibited AI practices are expected by February 2025, rules for general-purpose AI models by August 2025, while the rest of the obligations (including those on high-risk AI systems) will apply by August 2026.

The AI Act’s mission is commendable, as it wants to ensure AI is used in a safe and ethical manner. Commissioner for the Internal Market, Thierry Breton, expressed optimism about the Act’s potential to foster a successful startup ecosystem: “European democracy has delivered an effective, proportionate and world-first framework for AI, tackling risks and serving as a launchpad for European AI startups”. However, during its making, the AI Act has been heavily criticised from all sides, particularly regarding how it would overregulate certain AI models (e.g. GPAI models) and overburden smaller players, such as startups, with high compliance costs.

Recently, nobody other than Gabriele Mazzini, lead author of the proposal of the AI Act, voiced his concerns over the fact that the Act’s regulatory complexities would discourage European companies from investing in AI. The former Commission staffer added that the requirements, particularly for high-risk AI systems, could overwhelm startups, who often lack the resources to navigate complex regulations. For Mazzini, the situation is even worse when taking into account the EU’s lack of knowledge regarding how startups operate.

The AI Office: The startups’ last chance?
Against this grim background, what shall startups do? Well, it seems like the startups’ only hope is now the AI Office. Officially unveiled in May 2024, the European Commission’s AI Office will play a key role in the implementation of the AI Act, especially in relation to general-purpose AI models. It will also work to foster research and innovation in trustworthy AI and position the EU as a leader in international discussions

As of now, the Office has opened a call for expression of interest to participate in the drawing-up of the first GPAI Code of Practice and has launched a multi-stakeholder consultation on trustworthy GPAI models. This could be the startups’ final chance to have a say in the AI Act’s implementation process. Indeed, this first Code of Practice will detail the AI Act rules for providers of GPAI models. To demonstrate compliance, startups that are GPAI model providers will then be able to rely on the Code. For startups, making good use of these opportunities is not to be taken lightly, as the compliance costs remain very high. In the same vein, the AI Innovation Package, introduced to support AI startups, shall also be used to the best extent.

Conclusion
In the upcoming months and years, the AI Act’s implementation is expected to significantly impact the startup ecosystem. For this reason, startups (and all other relevant stakeholders) should use the AI Office as a way to ease the compliance process as much as possible. As other legislative initiatives are underway in the EU, such as the AI Liability Directive and a Directive on AI in the workplace, the startup ecosystem ought to consider all the options that can help startups thrive, grow, and innovate.