The EU AI Act presents significant challenges for companies, particularly small and medium-sized enterprises (SMEs). This article explores the implications of the act for HR departments using AI tools, outlines the compliance requirements, and provides insights into how businesses can navigate the evolving regulatory landscape.
Imagine you are the HR manager of a medium-sized manufacturing company, employing 250 individuals across Europe and North America. Your inbox overflows with resumes for each job opening, frequently exceeding 500 for a single position. The small HR team struggles to conduct thorough reviews, prompting the implementation of an in-house CV screening system powered by artificial intelligence.
This tool, built on a publicly accessible, open-weight foundation model and further refined using data from past successful hires, aims to identify promising candidates by evaluating their skills, experience, and overall fit. However, this seemingly beneficial tool has thrust your company into a complex regulatory landscape.\Under the EU AI Act, which officially came into force on August 1, 2024, but with staggered compliance deadlines, this fictional CV screener is classified as a high-risk application of AI. Other examples of high-risk applications include AI systems used for evaluating creditworthiness, AI managing critical infrastructure like railways and roads, and AI systems that determine access to education. These systems will be subject to stringent compliance requirements beginning on August 2, 2026. Non-compliance could lead to substantial financial penalties, potentially reaching up to 7% of a company's annual turnover. What once seemed like simple efficiency tools have now become potential liabilities. The AI Act applies to companies of all sizes that develop, sell, or utilize any type of AI system within the European Union, or whose AI system outputs are used in the EU. Furthermore, the General-Purpose AI Code of Practice, finalized in July 2025, provides concrete compliance obligations, based on computational thresholds rather than company size. Given the EU's regulatory influence, often referred to as the Brussels effect, these rules are poised to shape AI governance on a global scale. In a telling example, OpenAI recently advised California's Governor Gavin Newsom to treat developers of frontier models as compliant with state requirements if they align with a framework like the EU’s AI Code of Practice.\For small and medium-sized enterprises (SMEs), the challenges posed by the AI Act are particularly pronounced. SMEs often lack the resources of larger companies to adapt and comply quickly. The new rules could force SMEs to outsource AI compliance and innovation to expensive intermediaries. Furthermore, if the regulatory burden becomes too onerous, they might be discouraged from implementing AI tools altogether, which could lead to a loss of potential productivity gains. However, there's a path forward for SMEs. They need to actively implement specific strategies to navigate these challenges. The most strategic SMEs may even be able to use compliance with the AI Act as a means of differentiating themselves from competitors who are less prepared. SMEs must understand the rapidly evolving regulatory environment. While most provisions for high-risk AI systems, including those used in HR, will be fully in effect on August 2, 2026, the regulatory framework is already being established. The act's enforcement follows a phased timeline: the prohibitions of certain AI systems came into effect in February 2025, and obligations for General-Purpose AI (GPAI) models came into effect in August 2025. A comprehensive Code of Practice for GPAI models has been published to help providers demonstrate compliance. In practice, every firm integrating a general-purpose AI model into its AI systems—whether a niche chatbot or a foundation model—must provide clear documentation and publish a summary of the training data to meet the requirements. The precise computation thresholds are still under intense debate. The commission currently defines 10²⁴ FLOPS as an “indicative criterion” for GPAI, with a higher standard for models deemed to pose a “systemic risk,” although these standards are likely to be adjusted frequently. Adding to the complexity, the commission acknowledged that it may postpone some AI Act obligations due to delays in the promised European harmonized standards. In early July 2025, numerous leading European companies publicly urged the commission to “stop the clock” on the most demanding AI requirements. Given this fluctuating situation, SME leaders should plan as though the 2026 deadline is still in place, while maintaining budget flexibility in case Brussels grants more time. They can be certain that their obligations under the law will be extensive. Before an AI system can be marketed or used in the EU, its provider must undergo a rigorous conformity assessment to verify that it meets a range of requirements, including those pertaining to data quality, transparency, human oversight, and cybersecurity. Companies must also establish continuous risk-management systems to mitigate potential harm to users and data subjects
United States Latest News, United States Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
Navigating Vaccine Recommendations: Flu and COVID-19 Shots for ChildrenAs the fall respiratory season approaches, conflicting guidelines on flu and COVID-19 vaccinations for children create confusion. This article clarifies current recommendations from various organizations, including the CDC and AAP, to help parents make informed decisions regarding their children's health.
Read more »
Navigating a Thanksgiving Sibling Reunion: Seeking Closeness, Avoiding ResentmentAn 84-year-old woman seeks advice on how to have a positive Thanksgiving reunion with her siblings, focusing on fostering connection and productive conversation while avoiding rehashing old family issues and resentments stemming from their upbringing.
Read more »
Apple Calls for Repeal of EU's Digital Markets Act, Citing User Impact and Compliance ChallengesApple has requested the European Commission to repeal the Digital Markets Act (DMA), expressing concerns about its impact on users and the challenges it poses to compliance. The company highlights delayed feature rollouts, increased security risks, and dissatisfaction with the DMA's effectiveness in promoting competition and consumer choice.
Read more »
Coaching Real Leaders: Navigating a Leadership ChallengeExecutive coach Muriel Wilkins coaches Jada, a leader experiencing a career transition, by clarifying goals, understanding feedback, and strategizing for success. Jada discusses her career journey, challenges in the tech sector, and her passion for leadership.
Read more »
KuCoin appeals FINTRAC decision, reaffirms commitment to complianceKuCoin, today announced that it has formally appealed a recent decision by the Director of the Financial Transactions and Reports Analysis Centre of Canada (FINTRAC)
Read more »
Navigating Relationships and Friendships as Life EvolvesA woman over 50 navigates the complexities of dating someone who previously dated a friend and the challenges of maintaining friendships with those who have more free time. Advice is provided on how to approach the dating situation and manage the time constraints of a busy life.
Read more »