AI unlocks future-proof entrepreneurs. Bad regulation crushes them.

AI is the most powerful leveling force entrepreneurship has ever seen. A founder can now do what used to take a team of 10.

Shopify merchants know this advantage, and it's not just increased productivity: Orders coming to their stores from AI search are up 13x year-over-year. Those orders have a 49% higher conversion rate than traditional search, and average order values are 14% higher. AI is forecasted to contribute $2.9 trillion in GDP growth by 2030 in the U.S. alone.

But most of the regulatory response to AI forgets about entrepreneurs. They’re rarely in the room when the rules of AI get written, so we’re making sure they’re heard.

AI gave entrepreneurs an exoskeleton. Bad regulation pours a layer of lead on top.

The cost of AI regulation falls on entrepreneurs

Globally, overlapping national and regional AI regimes are multiplying. The EU wrote the most prescriptive AI regulation on the planet and is already walking it back through its recent AI Omnibus proposal. They’re acknowledging that their framework is unworkable barely a year after adoption, just like they’re doing with GDPR. But the regulatory pileup is accelerating.

65% of American small businesses are worried about AI compliance costs, up 27% in a year. And 77% say limits on AI would directly hurt their growth. 95% say their business would face challenges complying with AI regulations.

Under laws already passed or proposed, a business using AI to make certain decisions needs to conduct risk assessments, produce documentation, notify customers, and repeat it all annually. For a solo founder who just wants to use AI for basic tasks, laws like these might cause them to opt out and miss the opportunity of AI entirely.

Compliance costs are a regressive tax on ambition. Every new AI law, however well-intentioned, adds an additional burden onto entrepreneurs trying to compete.

Shopify’s stance

Shopify exists to make entrepreneurship more common. That is the principal lens we apply to AI regulation. From it, the question on every proposed law is simple: does this make it easier or harder to start a business? Most of what is being written today fails that test.

AI adoption is exploding for entrepreneurs. Sidekick, our AI commerce assistant, has had hundreds of millions of conversations and usage is up 4x year-over-year. In our Sidekick conversation logs, many of the merchants leaning in hardest are the ones with the least to spend on alternatives. This is happening now. So the importance of getting regulation right is urgent, not theoretical.

Based on everything we know about entrepreneurs, we defined six principles that every government should consider to protect AI innovation and entrepreneurship.

1. Don't write new AI laws, clarify the laws that exist

Privacy laws, consumer protection statutes, competition rules, anti-fraud legislation… these already cover most of the harmful outcomes we all want to prevent. If gaps emerge, they should be addressed through targeted updates to existing laws, not new ones. Creating a parallel, AI-specific legal labyrinth adds complexity, delays, and legal costs. All of that pulls businesses away from building, and hits the smallest ones hardest.

Entrepreneurs shouldn’t need a compliance team just to use a new tool.

2. Address the risk, not the tool

Not all AI uses are equally risky. A merchant using AI to write product descriptions is not the same as a hospital using it to make medical diagnoses, or a court determining a prisoner’s sentence. Regulation should reflect these differences, but most AI legislation written today doesn’t. Instead, it layers compliance obligations across all uses and all operators with fixed-cost mandates like audits, registrations, and pre-deployment approvals regardless of whether the actual risk justifies it.

If you’re a merchant using Sidekick, for example, it’ll help you do things like analyze sales data, identify growth opportunities, build marketing campaigns, and answer customer questions. None of that carries an inherently high level of risk, so it shouldn’t be regulated as if it does.

Proportionate rules are more fair, and more likely to address real harms without crushing businesses in the process.

3. Protect builders who act in good faith

AI developers and deployers who invest in reasonable safeguards shouldn’t face open-ended liability for every downstream outcome. Fear of unfair consequences kills experimentation and consolidates power in the big businesses that can afford the legal risk. Governments should establish clear safe harbor protections: if you build with care, disclose problems honestly, and follow established standards, you should not be one edge case away from a lawsuit. The regulatory goal should be less harm, not more litigation. Safe harbors give entrepreneurs room to innovate responsibly without having to risk edge-case liability.

Entrepreneurs need to be able to experiment. That’s how innovation happens.

4. Support open standards

AI-driven commerce should be built on open, interoperable standards, not proprietary ecosystems that lock merchants into a single platform. Regulation that mandates specific technical solutions concentrates power instead of distributing it.

This is why Shopify co-developed the Universal Commerce Protocol (UCP) with Google, backed by Amazon, Etsy, Meta, Microsoft, Salesforce, Stripe, Target, and Wayfair. An entrepreneur’s store should be discoverable on ChatGPT, Copilot, Gemini, and whatever comes next without them having to navigate complex integrations or platform-specific rules.

Open standards lower barriers to entry, increase competition, and allow the gains from AI to reach further.

5. Don’t create more patchwork

In the U.S. alone, state legislatures introduced 1,100+ AI bills in 2025. Overlapping national and regional AI regimes are multiplying globally. Each new law—whether it’s a state-level mandate, a Canadian framework, or an EU regulation that doesn't interoperate with allies—adds complexity for businesses. Governments need to exercise restraint. Clarify and enforce existing laws and resist standalone AI statutes. Reject duplicative or jurisdiction-specific registries, certification schemes, and disclosure obligations that don’t interoperate across markets.

Shopify merchants sell into more than 175 countries. Cross-border commerce is the default for a modern entrepreneur, even the ones outside of urban centres—rural cross-border sales are up more than 340% since 2019.

Every jurisdiction-specific AI law forces entrepreneurs to navigate another compliance regime, probably with no legal team to do it for them. None of it makes their product better or their customer safer.

6. Allow AI talent to build from anywhere

AI talent is concentrating in a small number of markets that have clearer rules and lighter compliance burdens. Restrictive or unpredictable frameworks slow innovation and push founders, researchers, and skilled workers to relocate. Governments should focus on predictable, proportionate, and interoperable policy that lets talent build and scale with minimal friction. The goal is innovation happening anywhere, not funnelled into a handful of hotspots.

Shopify is remote-first (we call it Digital by Design). Our team works from wherever they live and not the other way around. We see more and more entrepreneurs building their companies this way, too. There is an innovative advantage to allowing really smart people to build and work from wherever they choose. The next great founder shouldn't have to set up shop in another country just to access clearer rules.

Restrictive AI frameworks will push out the builders.

We need to get this right

So here's the ask: Don't write new AI laws when existing laws cover the harm. Regulate the risk, not the tool. Protect the builders acting in good faith. Back open standards. Resist the patchwork. And let talent build from anywhere.

Don’t pour on lead that slows down the builders.

Free up entrepreneurs to use their AI exoskeleton, and they’ll do what builders do best: innovate, create jobs, and drive economies.