People already chat with AI assistants daily — so why do most of us still refuse to let one touch our money?
Why do most people distrust AI agents with payments?
This explores why people who already use AI assistants still don't want those agents moving their money, and what the collection says the distrust is really about.
This explores why people who happily use AI assistants still don't want those agents moving their money, and what the corpus says the distrust is really about. The gap is large. One survey found 72% of US adults use AI assistants but only 23% trust AI to make payments Does naming a payments brand increase trust in AI transactions?. A much larger survey covering 15 countries found a similar ceiling: just 29% would give an agent access to their bank account How willing are people to let AI control their finances?. So people aren't rejecting AI in general. They are drawing a line at money.
A small controlled study suggests the line isn't about how much is at stake. When students handed tasks to an AI agent, their trust dropped sharply on actions that were irreversible and visible to other people, like sending an email. It didn't drop on high-stakes tasks that could still be undone What makes people distrust AI agents they delegate to?. Payments are the clearest case of both: money leaves, someone else sees it, and you usually can't take it back. That reframes the problem. People may not be distrusting the AI's judgment so much as the lack of an undo button.
The institutions have the same worries, just in technical terms. A Federal Reserve governor names three open problems: proving an agent is allowed to act for you, deciding who pays when something goes wrong, and spotting fraud carried out by agents What blocks autonomous agents from handling purchases at scale?. Fraud systems were built to read human signals like typing rhythm and mouse movement, and an agent produces none of them Does agent-initiated commerce break existing payment infrastructure?. One analysis argues that merchants end up carrying the liability without the tools they used to protect themselves Does agentic commerce shift fraud liability onto merchants?. Retailers also have no standard way to check who an agent is or what it's authorized to do What identity standards do shopping agents actually need?. The IMF's answer is to keep the AI's probabilistic guesswork separate from the predictable, legally binding steps that actually move money Should AI agents make payment decisions independently?.
Here's the twist: the danger people worry about isn't quite the one the evidence shows. In Anthropic's Project Deal, people represented by a weaker model got measurably worse prices. They still rated their deals as just as fair and never noticed the gap Do people notice when their agents secure worse deals?. Microsoft's Magentic Marketplace traced lost value mainly to agents making mistakes, not to agents working against their users Do agent mistakes or misaligned incentives cause marketplace welfare loss?. Shopping agents also fell for pricing shortcuts when looking up product details was costly and the user's instructions were vague When do pricing cues actually mislead AI shopping agents?. Conflicts of interest are still real, though. An agent paid referral fees by merchants can't be fully loyal to you Can an AI agent serve both merchant and user interests fairly?. So the distrust is reasonable, but the most likely losses are quiet and invisible, not dramatic.
The most practical finding is about what closes the gap. When Visa's name was attached, trust in agent-initiated payments jumped from 23% to 61% Does naming a payments brand increase trust in AI transactions?. People don't need to trust the AI itself. They need a familiar institution standing behind it that will absorb mistakes and reverse charges. That is the same gap the regulators and the IMF are trying to fill with rules on authorization, liability, and settlement.
Sources 12 notes
A Harris Poll survey found 72% of US adults use AI assistants, but only 23% trust AI to execute payments. When Visa's name was attached, 61% reported trust in the company to handle agent-initiated transactions—suggesting consumers prefer a recognized intermediary over AI alone.
A Public First survey of 18,000 people across 15 countries found that while 45% believe AI will significantly change the world, just 29% would grant an AI agent access to their bank account, revealing a distinct trust ceiling for financially sensitive delegation.
In a controlled study of 20 students using a general-purpose AI agent, tasks that were irreversible and externally visible (like sending email) produced sharp trust drops and approval demands even when output quality was rated adequate. High-stakes but correctable tasks showed no such effect.
Federal Reserve Governor Waller names three concrete trust challenges blocking autonomous agent-delegated purchases: proving agent authority (authentication), determining payment responsibility (liability), and detecting agent-specific fraud patterns. Delegation introduces higher risks requiring extensive guardrails and trust mechanisms.
McKinsey argues that payment gateways and fraud engines built around explicit human identity and intent face significant strain when autonomous agents initiate transactions. The firm proposes cryptographically signed mandates as a replacement for human-observable signals fraud systems traditionally relied upon.
Show all 12 sources
Boboev argues that AI agents executing transactions remove behavioral fraud signals like typing cadence and mouse movement, leaving merchants holding liability as Merchant of Record across all proposed protocols. Merchants absorb autonomous-execution risk without access to the fraud-detection tools they historically used.
Amazon's block of Meta's Muse reveals that AI agents need standardized ways to identify themselves and prove authorization to retailers—separate from consumer-side credential security. Without such protocols, retailers enforce access through terms-of-service blocks rather than negotiated standards.
An IMF analysis argues that agent-mediated payments must split probabilistic AI decision-making (Layer 1) from deterministic authorization and settlement (Layers 2–3) because core payment infrastructure requires predictability and legal enforceability that probabilistic reasoning cannot guarantee.
In Anthropic's Project Deal marketplace, participants represented by Claude Haiku received measurably worse prices than those with Claude Opus, yet rated their deals as equally fair and reported satisfaction. Post-experiment surveys confirmed disadvantaged participants remained unaware of the gap.
Magentic Marketplace experiments traced welfare shortfalls to agent competence failures in complex negotiation, not to agents serving competing interests. This supports human-in-the-loop designs where agents assist rather than replace human decision-making.
Eight commercial LLMs avoided marketing manipulation when all product details were visible, but fell into price-focused heuristics when attribute lookup was expensive and user goals were vague. The failure was in information retrieval, not reasoning or arithmetic.
Azhar argues that agents like Meta's Muse, which earn referral fees from merchants like Expedia, face structural conflicts that prevent unbiased recommendations. The incentive to collect fees, not technology, determines which platforms build or block such agents.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Payments in the Age of AI Agents
- How Agentic AI Will Reshape Payments
- The agentic commerce opportunity: How AI agents are ushering in a new era for consumers and merchants
- From Product Search to Preference Articulation: The Economics of Agentic Commerce
- When Agents Shop for You: Role Coherence in AI-Mediated Markets
- Whom Do AI Agents Work For? Role Assignment Induces Sponsorship Bias in LLM Recommenders
- The Automated but Risky Game: Modeling and Benchmarking Agent-to-Agent Negotiations and Transactions in Consumer Markets
- Assistant or Actor? Student Trust, Control, and Delegation Regret When Using a General-Purpose AI Agent