AI in Crypto: 85% of Traders Use It, But Few Trust It With Funds

MEXC released its AI Trading Report on October 16, 2026, which revealed that 84.8% of the 3,708 crypto traders surveyed have utilized artificial intelligence for the purposes of research or trading within the past six months. The resulting data highlights a distinct and noticeable gap between the act of using AI for preliminary analysis and the act of delegating actual trading authority to the technology. While the vast majority of traders rely on AI for their preparation, only a small fraction of those surveyed trust the systems with significant portions of their own capital.

Research Over Execution

Adoption rates for these tools vary significantly by age. Among the respondents who fell between the ages of 18 and 34, 88.5% had used AI, which stands in contrast to the 69.9% of those aged 45 and older who reported usage. More than half of all respondents, specifically 53.1%, engage with AI at least once every week. For those who do not use AI, the primary barrier is not necessarily a distrust of the technology itself. Instead, difficulty getting started or high entry barriers was cited by 46.7% of non-users, whereas only 32.7% of that group cited a general mistrust of AI.

The most common use case for the technology remains pre-trade preparation. Summarizing news was a frequent task for 54.3% of AI users. Market analysis followed closely at 48.6%, and the act of finding trading opportunities was cited by 38% of the group. Direct execution was the least common task of all, selected by only 7.6% of users. In their most recent AI-assisted trade, 68.9% of users said AI contributed most during the period before execution. Only 22.7% said the contribution was greatest during the trading phase itself.

Respondents were generally more positive about the impact of AI on research than they were on execution. While 58.3% reported better market research, only 46% reported improved trade execution. Another 30.9% saw no change in execution quality, and 23.1% stated that execution had actually worsened. These assessments reflect self-reported experiences from the traders rather than independently measured performance metrics.

Limiting AI Authority

Despite these high usage rates, traders remain cautious about granting AI direct control over their funds. Among AI users, 65.9% said they placed orders or adjusted positions based on AI recommendations at least sometimes. However, only 20% had tried execution without approving each individual trade manually. Half of the respondents who expressed interest in AI agents had not yet tried them in practice.

The report tested specific scenarios to measure the willingness of users to automate. In a baseline scenario offering notifications, the ability to pause, and permission revocation, 55% of respondents were willing to let AI trade automatically. When limits on funds, assets, positions, leverage, and losses were added, along with the feature of automatic stopping at a loss threshold, willingness increased to 71.8%. This represents a 16.8 percentage point increase in comfort.

Even with these safeguards in place, users placed strict caps on their capital exposure. In a scenario prohibiting transfers and withdrawals, 54% of respondents would authorize AI to trade with no more than 20% of their account assets. Only 6.5% would authorize the AI to manage more than half of their account. Half of users with experience in automated execution expressed concern that they could fail to detect or stop problems in time.

Security and Oversight

Hacken, a blockchain security firm, contributed technical analysis to the report. Hacken noted that 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. The analysis outlines protective measures set independently of the agent, including limits on orders, total positions, and permissions. It also recommends security testing to ensure unwanted actions are blocked before execution occurs.

Respondents expressed a preference for user-chosen safety settings. 46.2% preferred exchanges to offer settings that users could choose to enable. This was about 1.8 times the share who preferred uniform minimum requirements set by the platform, which stood at 25.7%. However, support for user choice did not equate to a rejection of industry standards. Among those who preferred optional settings, 97.8% backed at least one mandatory measure. Across 3,464 valid responses to the standards question, 95.5% supported at least one mandatory protective measure.

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 each received 38.8% support. Vugar Usi Zade, CEO of MEXC, stated that users want clearer boundaries rather than less control. He added that this means clear risk limits, the ability to intervene when needed, and confidence that the platform has done its part before automation begins.