Mutsamudu, Comoros, October 16, 2026 — MEXC, a global cryptocurrency exchange, released its AI Trading Report, based on a survey of 3,708 crypto traders. The report examines how respondents use AI, how much trading authority they would delegate, and what protective measures they expect from platforms. Blockchain security firm Hacken contributed technical analysis of risks associated with AI agent trading and the controls designed to address them.
Key takeaways:
- 84.8% of surveyed crypto users have used AI for research or trading. While 58.3% report improved market research, 46% also report better trade execution, though 54% have yet to see improvement at the execution stage.
- Interest in AI agents is already significant: half of respondents are interested in AI agents but have not tried them.
- 65.9% of AI users act on its advice at least sometimes; only 20% have tried execution without approving each trade.
- 54% would let AI trade with some funds, but no more than 20% of their account; only 6.5% would authorize more than half.
- Half of users with automated execution experience worry they could fail to detect or stop problems in time.
- Willingness to use automated trading grows from 55% under the baseline scenario to 71.8% when limits on funds, positions, leverage and losses are added.
- 95.5% support at least one mandatory protective measure, even as many want control over their own settings.
The report found that 84.8% of respondents had used AI to analyze or trade digital assets in the previous six months, with 53.1% using it at least weekly. Adoption was higher among younger participants: 88.5% of respondents aged 18–34 had used AI, compared with 69.9% of those aged 45 and older. Among respondents who reported not using AI, difficulty getting started or high entry barriers was a more common obstacle than AI mistrust, cited by 46.7% and 32.7%, respectively.
AI’s biggest role remained preparation before a trade. Summarizing news was among the most frequent tasks for 54.3% of AI users, ahead of market analysis at 48.6% and finding trading opportunities at 38%. Only 7.6% selected direct execution as one of their most-used tasks. Looking at their most recent AI-assisted trade, 68.9% said AI contributed most before execution, compared with 22.7% during trading.
Users were also more positive about research than execution. While 58.3% reported better market research, 46% reported improved execution. Another 30.9% saw no change in execution, and 23.1% said it had worsened. These assessments reflect respondents’ experiences rather than independently measured trading performance.
AI advice nevertheless influenced trading activity. Among AI users, 65.9% said they placed orders or adjusted positions based on its recommendations at least sometimes. These groups could overlap.
Risk limits changed respondents’ willingness to automate. In a scenario offering notifications, pausing and permission revocation, 55% were willing to let AI trade automatically. Adding limits on funds, assets, positions, leverage and losses, with automatic stopping at the loss threshold, brought willingness to 71.8% — an increase of 16.8 percentage points.
That willingness did not extend to most of an account’s funds. In a separate scenario prohibiting transfers and withdrawals and allowing users to stop trading at any time, 54% would authorize no more than 20% of their account assets and only 6.5% would authorize more than half.
Hacken’s contribution examines why withdrawal restrictions alone cannot prevent losses. An agent with trading access could still be manipulated into buying an asset at an inflated price or taking on excessive exposure. Its analysis outlines protective measures set independently of the agent, including limits on orders, total positions and permissions, alongside security testing to ensure unwanted actions are blocked before execution.
On oversight, 46.2% of respondents preferred exchanges to offer safety settings users could choose to enable, about 1.8 times the share who preferred uniform minimum requirements set by the platform (25.7%). Support for user choice did not mean rejecting industry standards: among those who preferred optional settings and answered the separate standards question, 97.8% backed at least one mandatory measure.
Across all 3,464 valid responses to the standards question, a risk assessment or knowledge test before first use received the most support at 55.3%. Independent audits and limits on capital and loss that cannot be disabled followed at 38.8% each.
“As AI becomes more integrated into trading, our findings show that users don’t want less control. They want clearer boundaries,” said Vugar Usi Zade, CEO of MEXC. “For traders, that means clear risk limits, the ability to intervene when needed, and confidence that the platform has done its part before automation begins.”
The full AI Trading Report, including Hacken’s security contribution, is available here.
About MEXC
Founded in 2018, MEXC is a leading global multi-asset trading platform built as your 0-fee gateway to infinite opportunities. Serving users across 170+ markets, MEXC provides simple and efficient access to crypto, stocks, tokenized assets, derivatives, and a growing range of TradFi-linked opportunities through one account and one gateway.
With 0 trading fees, deep liquidity, broad asset coverage, and a high-performance trading experience, MEXC is designed for retail users who want to discover earlier, act faster, and trade with fewer barriers. As crypto and traditional finance continue to converge, MEXC is committed to making global opportunities more accessible, helping users trade freely and MEXCmize every opportunity.
MEXC Official Website｜ X ｜ Telegram ｜How to Sign Up on MEXC
For media inquiries, please contact the MEXC PR team: media@mexc.com