About Services Consulting Workshop Certificates Events Work Blog Links Contact 🇵🇦 Español Book 10 free minutes →
← All articles
June 27, 2026 · Updated October 9, 2026 · 6 min read · By Chiragx

AI law for small business: what applies and what to do in 2026

AI lawAI regulationAI for businessEU AI Act
AI law for small business: what applies and what to do in 2026

For years, AI regulation felt like a problem for Google and OpenAI, not for your business. That changed in 2026. The rules apply by what your AI does, not by how big your company is, and a ten-person business that serves customers with a chatbot already has obligations.

This is the plain version, updated in October 2026, of what applies to you and what to do. No legal jargon.

What changed since June

  • Europe kept its date. Since August 2, 2026, the transparency part of the EU AI Act applies: if your AI talks to people, you must tell them it is an AI.
  • Europe postponed the heavy part. The "high-risk" obligations (AI used to hire people, grant credit or assess students) moved to December 2, 2027.
  • Colorado rewrote its law before it started. The 2024 law never took effect; the new, narrower version starts January 1, 2027.
  • I corrected a figure from the first version: the 7 percent of revenue fine is for prohibited practices, not for a chatbot that fails to disclose. The correct figures are below.

The EU AI Act: what already applies to you

It applies to any company whose AI produces output that is used in the European Union, wherever that company is based. Being outside Europe is no exemption if you serve European customers.

  • Disclose that it is an AI. Since August 2, 2026, whoever offers a chatbot or an assistant must tell people they are talking to an AI, at the latest at the first interaction, unless it is obvious.
  • Disclose deepfakes. If you publish AI-generated images, audio or video that look like real people or events, you must say they are artificial. Technically marking generated content is the job of whoever builds the tool, not yours.
  • Train your team. Since February 2025, anyone using AI in their business must take measures so their people know how to use it. There is no exam and no minimum level.
  • What is banned is already banned. Also since February 2025: manipulative AI, social scoring and emotion recognition at work or at school, among others.

The real fines

  • Prohibited practices: up to 35 million euros or 7 percent of worldwide annual turnover.
  • Almost everything else, including the duty to disclose an AI: up to 15 million euros or 3 percent.
  • Giving false or misleading information to authorities: up to 7.5 million euros or 1 percent.

For small businesses and startups, the lower of the two figures applies. In practice, the ceiling for a small business whose chatbot does not disclose it is an AI is 3 percent of its annual turnover. Each EU country sets the fine within those caps.

United States: Colorado and New York

  • Colorado. It passed the first broad state AI law in 2024, but rewrote it in May 2026 before it took effect. The new version starts January 1, 2027 and covers AI that materially influences decisions about employment, lending, housing, insurance, health care, education or legal services. You must give notice before using it, explain an adverse decision, let people correct their data and offer human review. Fines reach 20,000 dollars per violation, and only the attorney general can enforce them.
  • New York City. If you use an automated tool to screen candidates for a job in the city, you need an independent bias audit less than a year old, a published summary, and notice to candidates 10 business days in advance. There is no employee minimum.

What about Panama and Latin America?

  • Panama does not have an AI law in force yet. The National Assembly passed bill 588 on AI governance in August, but the president partially vetoed it on September 28, 2026 and it went back to the Assembly. What does apply today is Law 81 of 2019 on personal data protection, in force since March 2021 and supervised by ANTAI: if your AI uses customer or employee data, you need consent or a legal basis, and every person can ask for access, correction, deletion or objection.
  • Peru already has rules: the regulation of its AI law (Supreme Decree 115-2025-PCM) has applied since January 2026.
  • Brazil, Chile, Colombia and Mexico are still debating their laws; none is in force.

Where small businesses actually get caught

You do not need to be an AI company to be covered. The risk shows up in ordinary places: a chatbot on your site or on WhatsApp that does not say it is an AI, a tool that helps you screen job applicants or approve customers, or a model making decisions about pricing, credit or insurance. If AI touches a decision about a person, a rule probably applies.

How to apply it in your business

  • Disclose your bots. One sentence up front: "You are chatting with an AI assistant." It is the fastest way to get compliant.
  • Write a one-page AI policy. Which tools you use, for what, which data never goes in (passwords, health data, customer information without permission) and who reviews the output.
  • Map where AI decides about people. Hiring, credit, pricing, eligibility. There, always leave a human with the final word, and keep records.
  • Check for bias in any hiring or credit tool before you use it.
  • Take care of personal data. Consent, a clear purpose and a simple way for customers to ask to correct or delete their data.
  • Train your team. One practical session on what to do and what not to do with AI already counts.

Frequently asked questions

Does the EU AI Act apply to my small business if I am outside Europe?

Yes, if what your AI produces is used in the European Union: for example, a chatbot that serves European customers. If you only sell locally, it does not apply today, but the good practice is the same.

What does a small business need to do to comply?

The basics fit in one afternoon: disclose that the bot is an AI, write a one-page policy, keep a human on decisions about people and train your team.

Do I need an AI policy?

No Panamanian law requires one today, but it is the cheapest way to show you use AI carefully, and it is the fastest answer when a customer or a bank asks how you use it.

How much is the fine if my chatbot does not disclose it is an AI?

In the EU, up to 15 million euros or 3 percent of worldwide annual turnover. For small businesses, the lower of the two figures applies.

Does Panama have an AI law yet?

No. Bill 588 was partially vetoed by the president on September 28, 2026 and went back to the Assembly. Meanwhile, Law 81 of 2019 on personal data applies.

Why this is an opportunity, not just a burden

Most of your competitors are ignoring this and will scramble later. Getting it right now is cheap, builds trust with customers and turns "we use AI responsibly" into a selling point instead of a liability.

This is part of how I help companies adopt AI: systems that are useful, but also disclosed and defensible. If you use AI anywhere near your customers or your hiring, see how my AI consulting for businesses in Panama works, or let's talk.

This is general information, not legal advice. If your case is complex, talk to a lawyer.

Sources

Chiragx
Written by
Chiragx · Chiragx Bhakta
AI & technology advisor · Founder of CryptoManji, BBR Tek, Bell N Desk and Nova Ignis
About me · LinkedIn

This is where I can help

Business consulting in Panama, without textbook theory
See the service →
AI consulting and automation for businesses in Panama
See the service →

Want to talk about your project?

Book an advisory call →
Newsletter

Get every article in your inbox

AI and business analysis, no hype. Only when I publish something worth your time.