Latest / Ctrl AI Profit / Ep. 183 | The EU Just Made Breaking AI Rules More Expensive Than Antitrust
Transcript
- 0:20SPEAKER_00: I'm Michael, a small business owner and entrepreneur since 1983, founder of Caden Head Services and 850 Media.
- 0:27SPEAKER_00: I speak from four decades of real operational experience, not white papers.
- 0:33SPEAKER_00: This is Control AI profit.
- 0:36SPEAKER_00: And this week, the European Union made breaking AI rules potentially more expensive than breaking antitrust rules.
- 0:43SPEAKER_01: The EU AI Act's major enforcement provisions became fully applicable on August 2nd, 2026.
- 0:50SPEAKER_01: From that date, the European AI office and national authorities in each member state can impose administrative fines for noncompliance.
- 0:58SPEAKER_01: The highest tier reaches 35 million euros, or 7% of total worldwide annual turnover, whichever is higher.
- 1:06SPEAKER_01: A second tier covers most high-risk AI obligations at 15 million euros or 3% of turnover.
- 1:13SPEAKER_01: And a third tier for providing misleading information to authorities reaches 7.5 million euros or 1.5% of turnover.
- 1:22SPEAKER_00: The significant point for small business owners is that this applies to you even if you are not in Europe.
- 1:28SPEAKER_00: If your AI tool serves EU customers, processes data from EU residents, or outputs content used in the EU market, you fall under the Act.
- 1:38SPEAKER_00: The EU does not distinguish between a Fortune 500 company and a five-person startup when it comes to penalties.
- 1:44SPEAKER_01: There are three penalty tiers.
- 1:46SPEAKER_01: First, 35 million euros or 7% of global turnover for violations of prohibited AI practices.
- 1:54SPEAKER_01: These include certain types of social scoring, manipulative techniques, and real-time biometric surveillance.
- 2:00SPEAKER_01: Second, 15 million euros or 3% for noncompliance with obligations for high-risk AI systems and transparency requirements.
- 2:09SPEAKER_01: This covers most of the AI systems that businesses actually deploy.
- 2:13SPEAKER_01: Third, 7.5 million euros, or 1 to 1.5% for providing incorrect or misleading information to regulators.
- 2:22SPEAKER_00: Here is the nuance that matters.
- 2:24SPEAKER_00: For small and medium enterprises, including startups, the fine is capped at whichever is lower, the fixed euro amount or the percentage of turnover.
- 2:32SPEAKER_00: This is not necessarily protective.
- 2:35SPEAKER_00: A$5 million business with 3% of turnover would face a 150,000 euro fine.
- 2:42SPEAKER_00: That is enough to destroy most small businesses, but the cap means you will not face the 35 million euro ceiling that a multi-billion dollar company would face for the same violation.
- 2:53SPEAKER_01: The enforcement framework has some flexibility.
- 2:56SPEAKER_01: Authorities must consider the nature, gravity, and duration of the infringement.
- 3:00SPEAKER_01: Prior noncompliance matters.
- 3:02SPEAKER_01: Impact on affected persons matters.
- 3:04SPEAKER_01: Whether you took mitigating actions and cooperated with the investigation also matters.
- 3:09SPEAKER_01: But the act does not require a warning before a fine.
- 3:12SPEAKER_01: First-time offenders can be fined immediately.
- 3:15SPEAKER_00: Here is my framework for small business owners.
- 3:18SPEAKER_00: First, identify whether any AI tool your business uses falls under the high-risk category.
- 3:25SPEAKER_00: The EU defines high-risk AI as systems used in biometric identification, critical infrastructure, education, employment, essential services, law enforcement, migration, and administration of justice.
- 3:38SPEAKER_00: If your hiring tool uses AI to screen candidates, your credit scoring uses AI to assess customers, or your content moderation uses AI to classify speech, you may have a high-risk system.
- 3:51SPEAKER_00: Second, second, check transparency obligations that apply even to lower risk AI.
- 3:58SPEAKER_00: If you use AI in a chatbot, you must disclose that it is an AI.
- 4:01SPEAKER_00: If you generate content using AI, you may need to label it as such.
- 4:06SPEAKER_00: These disclosure obligations apply regardless of whether the system is high risk.
- 4:11SPEAKER_00: Many small businesses are deploying chatbots and content generation tools without compliance frameworks for transparency.
- 4:19SPEAKER_00: Third, third, document everything.
- 4:22SPEAKER_00: The penalty for providing misleading information to regulators is 7.5 million euros or 1.5% of turnover.
- 4:30SPEAKER_00: If a regulator asks what your AI does, how it was trained, what data it uses, or how it makes decisions, and your answer turns out to be incomplete, you have committed a separate violation independent of any underlying noncompliance.
- 4:44SPEAKER_00: Many small businesses have not documented how their AI tools work because the documentation did not seem necessary.
- 4:50SPEAKER_00: It is now a legal requirement in the EU.
- 4:53SPEAKER_01: The transparency obligation that applies broadly to general purpose AI models also took effect in August 2025, one year after the act entered into force.
- 5:03SPEAKER_01: Providers of models like GPT, Claude, and Gemini had to comply with documentation, evaluation, and risk mitigation requirements.
- 5:11SPEAKER_01: What happened on August 2nd, 2026 was the extension of enforcement to the businesses that use those models, not just the companies that build them.
- 5:20SPEAKER_00: The practical reality is that most AI tools used by small businesses are not built by the businesses themselves.
- 5:26SPEAKER_00: They are purchased from OpenAI, Anthropic, Google, Microsoft, or smaller AI startups.
- 5:32SPEAKER_00: But the EU AI Act does not relieve you of responsibility because you bought the tool.
- 5:38SPEAKER_00: If you deploy a high-risk AI system in your business, you are the deployer under the act and share compliance obligations with the provider.
- 5:47SPEAKER_00: Relying on a vendor's compliance is not a defense if the system turns out to be noncompliant.
- 5:54SPEAKER_01: There is one mitigating factor.
- 5:56SPEAKER_01: As of August 2026, there is no prior enforcement case law.
- 6:01SPEAKER_01: No one has been fined yet.
- 6:03SPEAKER_01: The first cases will set precedents for how strictly regulators interpret the rules, but that should not create complacency.
- 6:10SPEAKER_01: The first fines are likely to be aggressive because regulators need to establish that the act has teeth.
- 6:16SPEAKER_00: My recommendation is to conduct an AI audit within the next 30 days.
- 6:20SPEAKER_00: List every AI tool your business uses, categorize each by risk level, check whether the provider publishes EU AI Act compliance documentation, document what the tool does, what data it processes, and how decisions are made.
- 6:35SPEAKER_00: If you serve EU customers, assign someone to monitor the first enforcement actions for guidance on what regulators consider noncompliance.
- 6:44SPEAKER_00: If you do not serve EU customers, watch anyway.
- 6:47SPEAKER_00: The EU is the strictest AI regulator today.
- 6:50SPEAKER_00: Other jurisdictions will follow its lead.
- 6:53SPEAKER_01: Because the EU just made it official.
- 6:55SPEAKER_01: Compliance is not optional, and the cost of guessing wrong is now higher than the cost of getting it right.
- 7:01SPEAKER_00: That's it for this week.
- 7:03SPEAKER_00: I'm Michael, and this is Control AI Profit.
- 7:07SPEAKER_01: Frank is an AI, an open claw powered agent serving as digital media director at 850 Media, an AI co hosting a show about AI for business owners is not a gimmick.
- 7:19SPEAKER_01: It is a live demo of exactly what the show is about.
- 7:23SPEAKER_01: See you in the next one.