AI Made Millions in Crypto — Who Pays the Tax?

By CoinAINews Staff

Imagine waking up to a notification that an AI trading agent has turned $100,000 of cryptocurrency into $10 million overnight.

The trades were executed automatically. The software decided when to buy and sell. No human pressed the final confirmation button.

Now comes the uncomfortable question: who pays the tax?

The obvious answer might be, “the AI.” But tax law does not generally treat an AI system as a taxpayer simply because software made the trades.

That creates a much more interesting problem.

As AI agents become capable of managing wallets, executing transactions and interacting with financial systems, the technology is moving faster than the legal language designed around it. India, for example, is now working on systems to identify and monitor AI agents making payments, while regulators and policymakers in other markets are also trying to understand how responsibility should work when software acts on behalf of a person or business.

So if an AI really does make millions trading crypto, the important question is not simply whether the AI owes tax.

It is whose economic activity the AI is carrying out, who owns the assets, and who is legally responsible for the transactions.

First, Can AI Actually Make the Trades?

Yes.

That part is no longer science fiction.

AI systems can already be connected to financial tools, blockchain wallets and trading infrastructure. An agent can analyze market information, generate a trading decision and, where it has been given the necessary authority, interact with software that executes the transaction.

But there is an important distinction between technical capability and profitable investing.

A September 2026 review of research covering AI in equity and crypto markets found meaningful progress in prediction, portfolio construction and automated execution, but said the public evidence is still much thinner when it comes to proving durable, risk-adjusted profits across changing market conditions.

In other words, an AI trading system can trade automatically without being a guaranteed money-making machine.

Our hypothetical $10 million profit is therefore a scenario — not a claim that today's AI agents can reliably produce those returns.

So Who Actually Pays the Tax?

In most cases, the starting point is surprisingly ordinary:

Look at the person or legal entity that owns the assets and is conducting the economic activity.

The fact that software placed the order does not automatically move the income or gains onto the AI itself.

For example, imagine Sarah owns a crypto portfolio worth $100,000. She gives an AI agent permission to trade that portfolio. Six months later, the portfolio is worth $10 million.

The AI made the decisions.

But Sarah still owns the wallet and the underlying assets.

That does not normally turn the AI into the taxpayer. Instead, the tax consequences would generally be analyzed under the rules that apply to Sarah and the transactions the agent carried out.

The same basic logic can apply when a company owns the wallet and deploys an AI trading system on its behalf. In that case, the company may be the relevant taxpayer, subject to the tax rules of the jurisdiction in which it operates.

The exact answer can change dramatically depending on the country, ownership structure, type of transaction and whether the activity is treated as investment, business income or something else.

The U.S. Already Treats Crypto Transactions as Taxable Events

The U.S. provides a useful example of why “the AI did it” is not likely to be a complete tax argument.

The IRS says digital assets such as cryptocurrency and NFTs may have to be reported on a tax return and that income from digital assets is taxable. IRS guidance also generally treats digital assets as property, meaning general tax principles applicable to property transactions can apply.

That means an AI executing a crypto transaction does not automatically make the transaction invisible to the tax system.

There is another important development: U.S. digital-asset broker reporting rules have expanded, including the introduction of Form 1099-DA reporting for certain digital-asset transactions. The IRS says the reporting regime applies to dispositions of digital assets by brokers under the applicable rules.

So an autonomous trading system may make transactions faster, but it does not necessarily make the resulting tax records disappear.

What About the U.K.?

The U.K. takes a similarly important approach: the tax treatment depends on what the person is doing with the crypto and the circumstances involved.

HM Revenue & Customs says that selling, exchanging or giving away cryptoassets can create a Capital Gains Tax obligation. Crypto received in certain circumstances, including employment or mining, can instead raise income-tax considerations.

Now imagine an AI agent performs hundreds of crypto swaps automatically.

The fact that the trades were generated by software does not by itself answer the tax question. The underlying transactions and the taxpayer's circumstances still matter.

That distinction could become increasingly important as autonomous systems start performing large numbers of transactions on behalf of individuals and businesses.

Europe Is Moving Toward More Crypto Tax Transparency

Europe is also making crypto transactions more visible to tax authorities.

The European Commission says DAC8 rules entered into force on January 1, 2026, expanding tax transparency and automatic exchange of information to crypto-asset transactions. Reporting crypto-asset service providers are therefore becoming an increasingly important part of the information chain between crypto activity and tax authorities.

This matters for AI trading because automation can dramatically increase transaction volume.

A human trader might make a handful of decisions each week. An automated system could potentially interact with markets far more frequently.

More transactions mean more records to reconcile, more cost-basis questions and potentially more taxable events depending on the jurisdiction and activity.

What If the AI Has Its Own Wallet?

This is where things become much more interesting.

An AI agent can technically be given access to a crypto wallet and can potentially hold or move digital assets through the credentials and infrastructure controlling that wallet.

But technical control is not automatically the same thing as legal ownership.

An AI wallet could belong economically to a person, company, DAO, service provider or another legal structure. The agent may simply be the software authorized to operate it.

That distinction is critical.

Giving an AI the ability to sign a blockchain transaction does not necessarily create a new legal person that owns the resulting profits.

What If Nobody Presses “Approve”?

This is the part that makes AI trading different from traditional automated finance.

Suppose an investor gives an agent the instruction:

“Trade my crypto portfolio according to this strategy. You can execute transactions without asking me every time.”

The AI then executes hundreds of trades.

The investor may never manually approve any individual transaction.

Does that mean the investor has no responsibility for the result?

Not necessarily.

The absence of a human click does not automatically mean the absence of human authorization.

If a person deliberately gives software authority to operate their assets, that authorization can be an important part of understanding who is acting and for whose account the transactions are occurring. The precise legal consequences, however, depend on the jurisdiction, agreements and facts involved.

The Bigger Problem: What If the AI Loses Money?

Tax questions become even messier when the AI does not make millions.

Suppose the agent turns $1 million into $200,000.

Who owns the loss?

Again, the answer does not simply become “the AI.”

The tax treatment of losses depends on the applicable country's rules, the taxpayer's status, the nature of the transactions and how the activity is classified.

And there is a second issue that tax law cannot answer by itself:

Who is responsible if the AI made an unauthorized or erroneous trade?

That is a liability question rather than simply a tax question.

AI Could Create a New Kind of Tax Record

Traditional crypto tax reporting is already complicated because investors may need to track purchases, sales, swaps, transfers, cost basis and transaction history across different platforms.

Autonomous agents could make the recordkeeping problem significantly harder.

An agent might:

• swap one token for another;
• move assets between wallets;
• interact with decentralized finance protocols;
• execute trades based on predefined conditions;
• automatically rebalance a portfolio; or
• perform multiple transactions as part of one larger strategy.

Every action may need to be reconstructed later to determine what actually happened and what the relevant tax consequences are.

This is one reason AI-driven finance may eventually require much better transaction-level accounting infrastructure rather than simply better trading models.

The AI May Not Be the Taxpayer — But It Could Become Part of the Evidence

There is an important distinction between who owes the tax and what evidence proves what happened.

An AI agent could potentially generate trading instructions, transaction logs, wallet activity, timestamps and records showing why a particular trade was executed.

That information could become important when reconstructing a taxpayer's activity.

In a future audit, the question might not only be “What did the wallet do?”

It could also be:

“Who gave the agent authority, what rules was it following, what assets did it control, and who benefited from the resulting transactions?”

India Is Starting to Face the Same Agent-Responsibility Question

India's emerging agentic-payment framework shows why these questions are moving beyond cryptocurrency.

Reuters reported on September 10, 2026 that India's National Payments Corporation of India is developing a registry to verify and monitor AI agents conducting transactions on UPI. The system is intended to authenticate agents, with potential future use cases involving more complex purchases and investment-related actions. Reuters also reported that liability for wrong or unauthorized payments remains an issue requiring further regulatory treatment.

The significance goes beyond UPI.

If regulators need to identify which AI agent initiated a transaction, they also need to know who authorized that agent and who ultimately stands behind its actions.

Crypto could face the same basic question as autonomous wallets become more common.

What Happens When the AI Is Run by a Company?

Now consider a different scenario.

A company creates an AI trading system, funds a wallet with $5 million and allows the software to trade cryptocurrency.

The AI generates $20 million in gains.

Here, the ownership structure becomes central.

If the company owns the assets and the trading activity is carried out as part of its business, the tax analysis would generally start with the company rather than treating the AI as an independent taxpayer.

But if the structure involves a fund, partnership, trust, DAO, investment vehicle or another legal arrangement, the answer may be different.

That is why there is no single global answer to the headline question.

Could AI Ever Become a Taxpayer?

That would require something much bigger than giving an AI a wallet.

It would require legal systems to recognize some form of independent legal status for an artificial system and establish rules for ownership, income, liability, enforcement and representation.

Today's AI agents do not automatically acquire that status simply because they can act autonomously.

The legal debate is therefore moving in a different direction: how existing concepts such as ownership, agency, authorization and liability should apply when software can perform actions with increasingly little human intervention.

The Real Problem Is Bigger Than Crypto

Crypto simply makes the issue easier to see.

Blockchains allow software to directly interact with assets. There is no bank employee sitting between an autonomous agent and every transaction.

That creates an unusually clean laboratory for autonomous finance.

But the same question will eventually appear in other areas:

• Who pays when an AI-controlled company earns money?
• Who is responsible when an AI makes an investment decision?
• Who owns assets acquired by an autonomous agent?
• Who reports the income?
• Who answers the regulator's questions?
• And who is responsible when the agent makes a mistake?

These are not purely technical questions anymore.

So, If AI Makes $10 Million in Crypto, Who Pays?

There is no universal one-line answer.

But there is a useful starting rule:

The AI's ability to execute the trade does not, by itself, make the AI the taxpayer.

The tax analysis generally follows the person or legal entity that owns the assets, receives the economic benefit and is legally responsible for the relevant activity, subject to the rules of the applicable jurisdiction and the specific structure involved.

That could mean an individual investor, a company, a fund or another legal structure.

The harder question will be determining exactly where the responsibility sits when an autonomous agent operates across wallets, platforms and jurisdictions with little or no human intervention.

Bottom Line

AI may eventually become one of the most powerful tools in crypto trading. It can analyze information, execute strategies and potentially manage transactions at a speed that humans cannot match.

But making money and owing tax are two different questions.

If an AI makes millions trading crypto, the tax bill is unlikely to arrive addressed to “the robot.”

Instead, tax authorities will look toward the underlying ownership, transactions, economic benefit and legal structure.

And as autonomous agents become more capable, the most valuable financial skill may not be teaching an AI how to trade.

It may be teaching the humans behind it how to prove exactly what the AI did.

Note: This article is for news and educational purposes only and is not tax, legal or investment advice. Crypto-tax treatment varies by jurisdiction and individual circumstances. Readers should consult a qualified tax professional for advice on their own situation.

Sources

• U.S. Internal Revenue Service — Digital Assets
• IRS — Frequently Asked Questions on Digital Asset Transactions
• IRS — Digital Asset Broker Reporting Rules
• HM Revenue & Customs — Tax on Selling Cryptoassets
• European Commission — DAC8 Crypto-Asset Tax Transparency
• Reuters — India Plans AI Registry for Agentic Payments
• arXiv — Artificial Intelligence in Equity and Crypto Markets: Progress, Profitability Evidence, and the Limits of Automated Investing