The EU AI Act’s February Compliance Wall refers to 2 February 2025 when the first and most critical provisions of the regulations take effect. Unlike later deadlines focused on documentation, this date bans certain high-risk AI practices and requires immediate staff training.
Below is a summary of the requirements effective 2 February 2025:
Key prohibitions (unacceptable risk)
Practices posing an unacceptable risk to safety, livelihood, or rights are now prohibited (Chapter 2, Article 5).
- Manipulative AI: systems using subliminal or deceptive techniques to distort behavior materially.
- Exploitation of vulnerabilities: Systems taking advantage of vulnerabilities such as age, disability, or socio-economic status to cause harm or distort behavior.
- Social scoring AI systems used to evaluate, classify, or score individuals based on social behavior, possibly resulting in detrimental treatment.
Biometric Surveillance & Categorization
- Real-time remote biometric identification in public places for law enforcement erupts in narrowly defined cases.
- Un-targeted collection of internet images or CCTV footage to build or expand facial recognition databases.
- Emotion recognition systems in workplaces and educational institutions.
- Biometric categorization to infer sensitive characteristics such as race, political opinions, or sexual orientation.
Predictive policing: AI systems assessing the risks of criminal offenses based solely on profiling or personality traits.
Mandatory AI Literacy (Chapter 1)
Companies that provide or deploy AI systems must ensure adequate AI literacy among staff and others who handle these systems on their behalf.
- Action required: Organizations must promptly provide training to ensure staff understand AI opportunities, risks, and potential harms.
Sanctions for Noncompliance.
Sanctions for violations are severe and intended to deter non-compliance.
- Fines up to €35 million or 7% of total worldwide annual turnover from the previous financial year, whichever is higher.
Who is Affected?
The Act applies to:
- Provider’s entities developing AI systems, whether inside or outside the EU.
- Deployers: Entities using AI systems professionally within the EU.
- Importers/distributors: Entities placing AI systems on the EU market.
- Extra-territorial reach: Companies outside the EU must comply if their AI systems impact individuals in the EU.
Upcoming Compliance Deadlines
February 2025 compliance deadline marks the start of a staged rollout:
- 2 August 2025: Rules on general-purpose AI models and governance, including associated fines, take effect.
- August 2, 2026: Most remaining provisions of the AI Act, including rules for high-risk AI systems, become applicable.
- 2 August 2027: Compliance required for high-risk AI systems embedded in already regulated products, such as medical devices and toys.
In Brief
The EU AI Act establishes a comprehensive statutory framework for companies operating AI systems in the EU. From 2 February 2025, regulated companies are required to ensure AI literacy and avoid prohibited AI practices. Non-compliance can result in significant fines.
The Applicability of Chapters I and II of the AI Act
EU Regulation 2024/1689 (AI Act) establishes a uniform legal framework for the development, marketing, deployment, and use of artificial intelligence systems (AI Systems) in the EU. The act entered into force on 1 August 2024, with its rules taking effect at later dates. Chapters 1 and 2 will apply from 2 February 2025, and companies should prepare accordingly. Below is a summary of these chapters:
Chapter I : AI Literacy
Chapter 1 defines the scope of the AI Act and provides definitions. Article 4 requires companies that provide or deploy AI systems to ensure mandatory AI literacy within their organizations.
AI literacy refers to the skills, knowledge, and understanding that enable providers, deployers, and affected individuals to make informed decisions about AI systems and understand their opportunities, risks, and potential harms.
To comply with AI literacy requirements, companies have to ensure that their staff and others involved in operating or using AI systems have adequate AI literacy. This usually involves promptly organizing training and education for all relevant personnel.
Chapter 2: Prohibited AI Practices
Chapter on prohibited AI practices will also apply from 2 February 2025. The following practices will be prohibited from that date:
- By systems that use subliminal, manipulative, or deceptive techniques to intentionally or effectively distort behavior;
- AI systems that take advantage of vulnerabilities of individuals or groups to intentionally and effectively distort their behavior;
- AI systems that recognize emotions in workplaces or school environments; and
- AI systems that create or expand facial recognition databases using internet images or CCTV footage.
Non-compliance with rules on prohibited AI practices may lead to administrative fines of up to €35 million or 7% of the company’s global annual turnover. Member states may also establish additional sanctions, including those for failing to meet AI literacy requirements.
