Yes. Copy-trading services, strategy platforms and pooled AI-run funds all let you follow someone else’s algorithm. But a January 2026 academic study found shared strategies face a bind: public ones get picked apart by faster traders, cutting one test strategy’s profits by 27.6%, while secret ones let insiders exploit followers. And if it goes wrong, who is responsible is still unsettled.
A note on the legal and regulatory points here: they summarise public sources as of November 2026 and should not be relied on. Rules for AI agents are still being written, and some of what is described here is proposed or coming soon rather than settled. Check the current position, or ask a qualified professional, before acting on it.
What does copying an AI strategy actually look like?
Copying an AI strategy means letting someone else’s trading logic decide your trades, either by mirroring their orders or by handing your funds to their system. The academic study calls these collective investment algorithms: trading logic shared by a community of users who all follow it.
It groups them into several types, from traditional funds to AI-run investment groups. For crypto users, they mostly show up in three forms:
| Form | What you hand over | What you usually see | The main catch |
|---|---|---|---|
| Copy-trading | Permission to place trades in your account that mirror a leader’s | The leader’s past results and current trades | You trade after the leader, at a worse price |
| Strategy platforms | Trade access to a subscribed algorithm | A description and a backtest, rarely the code | The logic can change without you noticing |
| Pooled AI-run funds | Your deposit, into a shared pot an agent trades | The pot’s balance and on-chain moves | You are one creditor among many, and insiders may know more |
In each case the AI part may be large or tiny. Some products are a simple rule-based bot sold with an “AI” label. Others hand real decisions to a model. Ask which one you are buying.
Why doesn’t a copied strategy perform like the original?
Because you are not running the original. You are running a delayed, larger, more visible copy of it, and each of those changes costs money.
Delay. A copier trades after the leader. In fast markets, the price has already moved by the time your order lands.
Size. A strategy that works with a small account may not work when thousands of followers pile in. Their combined orders move the price against everyone, including the leader.
Visibility. If other traders can see what the strategy will do, they can trade ahead of it. The academic study tested this directly. A bot that copied a transparent strategy’s trades one block ahead of it captured nearly all the profit the strategy would otherwise have made.
There is also the question of whether the original was any good. The headline evidence on AI trading is weak. In one open-ended trading contest reported in April 2026, the top human traders beat the top AI agents by more than five times. In another, only three of seven leading AI models turned a profit per trade.
A copy cannot be better than its source. And a public leaderboard shows the strategies that survived, not the ones that quietly closed. That last point is our inference about how leaderboards work, not a measured figure.
What is the transparency trade-off?
Every shared strategy has to choose between being open, which lets outsiders exploit it, and being secret, which lets insiders exploit you. The study’s authors call this the CoinAlg bind, and they conclude that no design fully escapes it.
If the strategy is public, arbitrage bots can read its next move and trade around it. In the study’s tests on a large decentralised exchange, sandwich attacks, where a bot buys just before a known order and sells just after, cut a transparent strategy’s profits by 27.6%.
If the strategy is secret, you must trust whoever runs it. The study found that even a tiny leak is enough. When a private strategy leaked only the direction of its next trade, ten blocks ahead, a frontrunner still cut its profits by 17.8%. Anyone with inside access could do the same, at followers’ expense.
The researchers tested 3,072 parameter combinations on two trading pools, using market data from June 2024 to July 2025. They conclude that shared strategies are here to stay, and that sound guardrails remain an open problem. That is an honest summary of where the field stands.
The practical lesson: when someone says their strategy is both fully transparent and highly profitable, ask how it survives the bots. When they say it is secret, ask who can see the trades before you do.
If a copied strategy loses money, who is responsible?
It depends on who runs the service, where it is based and what you agreed to, and for AI-run strategies the rules are still being worked out. The open questions are being asked at the top.
At a CFTC forum in October 2026, the CFTC chair asked publicly whether an agent’s developer becomes subject to registration, and who is liable if an agent goes rogue. In March 2026, the chair said the agency planned to address whether software providers trigger its registration requirements.
Existing rules still apply where they already reach. In December 2024, CFTC staff told regulated firms that their obligations under existing law continue when they use AI. The international body of securities regulators, in its 2023 DeFi report, said the people and firms behind arrangements described as decentralised typically remain responsible.
None of that tells you whether you can recover a loss from a strategy author. Our judgement: assume any loss is yours unless a named, regulated entity has accepted responsibility in writing. Read the terms before you connect anything.
What could go wrong?
The serious risks are less about the AI and more about access, incentives and crowding. Here they are, specifically.
Insiders trade ahead of you. In a private strategy, whoever sees the signals first can profit from them before followers execute. The study shows how little leaked information that takes.
Too much access. Copying usually needs permission to trade in your account. It should never need permission to withdraw. If a service asks for withdrawal rights, treat that as a reason to walk away.
Hidden leverage. A strategy that looks steady may be using borrowed money. A copied position can be liquidated faster than you can react.
The logic changes. The author can update a strategy at any time. What you vetted last month may not be what is trading today.
Everyone exits together. A popular strategy means thousands of followers holding the same positions. Our guide on what happens when thousands of agents run the same strategy explains why that turns a dip into a rush for the door.
Faster bots on the other side. Predictable followers are exactly what speed traders look for. Our guide on whether you are trading against humans or bots covers who is likely waiting.
The trust takeaway
Copying a strategy is not copying a result. You inherit the strategy’s risks, plus the costs of being late, being large and being visible. The right question is not “how much did it make?” but “who else can see my trades, and who answers if it breaks?”
Before you copy any AI trading strategy, check:
- Is the logic public or secret? If secret, who can see trades before followers do?
- What access does it need? Trade-only, with no withdrawal rights, should be the limit.
- How does it size your position? And how long after the leader do your trades execute?
- Is the track record live and verifiable? On-chain history beats screenshots and backtests.
- What are the fees? Include performance fees, subscription costs and trading costs.
- Who is responsible if it fails? A named entity, a jurisdiction and written terms, or nobody?