The European Union began enforcing major parts of its Artificial Intelligence Act on Sunday, August 2, giving regulators new powers to investigate advanced AI models and fine companies that break the rules.
The EU adopted the AI Act in 2024 and has introduced its requirements in stages. The law follows a risk-based system, meaning companies face stricter obligations when their AI systems pose greater risks to people’s safety, health or fundamental rights.
Chatbots and Deepfakes Must Be Clearly Identified
From August 2, companies must ensure that users know when they are interacting with an AI system such as a chatbot.
Providers must also make AI-generated or manipulated content detectable, while deepfakes and certain AI-generated text on matters of public interest must carry clear labels. Artistic, creative, satirical and fictional works receive limited exemptions from some disclosure requirements.
AI systems placed on the market before August 2 have until December 2, 2026, to meet the technical marking and detection requirements. Content created before August 2 does not need to be labelled retroactively.
The Act also prohibits practices including harmful AI manipulation, individual criminal-risk prediction, commonly described as predictive policing, and emotion recognition in workplaces and educational institutions. Most of these bans have applied since February 2025.
EU Gains Powers Over Advanced AI Models
The European Commission’s AI Office can now enforce rules covering general-purpose AI models, including advanced systems that power chatbots.
The office, which includes technology experts, lawyers and economists, can request documents, examine models, order corrective measures and require companies to provide access to their technology. Regulators can also restrict a model’s use in the EU if its provider fails to address identified risks.
The new powers follow difficulties the EU faced while seeking access to Anthropic’s Mythos model earlier this year. National authorities will remain responsible for enforcing rules covering smaller AI systems.
Companies Face Heavy Fines
Companies involved in prohibited AI practices can face fines of up to €35 million, or 7 percent of their annual worldwide revenue, whichever is higher.
Other breaches can result in fines of up to €15 million or 3 percent of global annual revenue. Regulators may also impose penalties when companies provide misleading information or fail to cooperate with an investigation.
More AI Rules Are Coming
From December 2, the EU will ban AI systems designed to generate non-consensual sexually explicit or intimate deepfakes, including so-called nudification tools. The change follows international concern over AI-generated sexual images, including content produced through Elon Musk’s Grok chatbot.
Meanwhile, the EU has delayed some requirements for high-risk AI systems to give companies more time to comply.
Rules covering stand-alone systems used in areas such as biometrics, education, employment, migration and critical infrastructure will apply from December 2, 2027. Requirements for AI built into regulated products, such as toys and lifts, will take effect on August 2, 2028.
The post EU Enforces World’s Toughest AI Rules With Up to €35 Million Fine appeared first on ProPakistani.
