The European Union has set a date of August 2, 2026 for European companies to end unregulated experimentation with Artificial Intelligence in their offices. With this ultimatum, the grace period for the adoption of the new technological requirements is officially over and the European Commission assumes the power to demand strict compliance under the threat of million-dollar fines.
After the initial phases that prohibited unacceptable practices, it is now the turn of developers and users of high-risk systems, who must get down to the task of auditing their algorithms to move towards a safer digital market against possible AI biases and failures.
The measure is part of the pioneering Artificial Intelligence (AI) Regulation, which regulates risk management, transparency and human supervision obligations across the continent. The text establishes that companies must adopt quality management systems, technical documentation and post-market monitoring.
The deadline for companies is August 2, 2026, at which time most of this regulation will be fully in force and the competent authorities will begin their surveillance and enforcement work. Starting this summer, the warning is clear: high-risk systems (such as those used in employment, education, justice or critical infrastructure) will have to meet strict requirements to operate.
The million-dollar fines and the “cascade effect”
Although Regulation (EU) 2024/1689 aims directly at developers, the regulations contemplate a total transformation of the productive fabric. The measures will require control throughout the entire algorithmic value chain. That is to say, thousands of small and medium-sized companies (SMEs), integrators and importers that implement or distribute these tools will be dragged by the new standard, which will have to raise their standards if they want to continue operating.
The regulation not only requires transparency, but introduces severe penalties. The law contemplates astronomical fines that can reach up to 35 million euros or 7% of the global turnover of the offending company. “Technical ignorance” will no longer be a valid excuse under the law, forcing implementers to assign competent and trained human supervision.
The EU will evaluate the impact through a new central command
The main objective of this regulation is to stop the enormous social impact that uncontrolled AI can have on the fundamental rights of Europeans.
In fact, to coordinate this colossal task, the European Artificial Intelligence Office (AI Office) is already carrying out its functions, a body attached to the European Commission that will centralize the supervision of general-purpose models, investigate possible infringements and may directly demand corrective measures from suppliers.
With this roadmap, Brussels seeks to completely transform the way we manage technological risks on a day-to-day basis, starting with a gesture as simple as forcing business management to guarantee human intervention and respond to the law if their algorithmic systems fail.