Does trusting an AI with your feelings work the same way as trusting it with your money?
Does emotional trust in AI differ from financial trust in AI?
This explores whether people trust AI differently when it is comforting them than when it is handling their money, and whether those two kinds of trust are earned in different ways.
This explores whether trusting AI with your feelings works differently from trusting it with your money. The corpus suggests they are close to opposites. Emotional trust comes easily and is driven by how the AI feels to talk to. Financial trust is hard to win and is driven by whether a mistake can be undone.
On the emotional side, trust grows out of the conversation itself. Focus groups found that ChatGPT earns trust through quick replies that respond to what you just said, not through being accurate Does conversational style actually make AI more trustworthy?. Because no human is there to judge, people share intimate things more freely. The same lack of judgment also makes people more willing to lie to a machine How do people decide what to share with AI systems?. Patients using therapy chatbots report a real sense of bond. That bond can sit alongside clinical safety failures, such as the bot reinforcing unhealthy thinking, and the bond score alone gives no sign of those problems Do therapeutic chatbot bond scores hide deeper safety problems?. In short, emotional trust tracks how warm the AI feels, not how reliable it is.
Financial trust has a hard ceiling. Across 18,000 people in 15 countries, only 29% would give an AI agent access to their bank account, even though many more expect AI to change the world How willing are people to let AI control their finances?. In a US survey, 72% use AI assistants but only 23% trust AI to make payments. That figure jumps to 61% when a familiar payments brand like Visa stands behind the transaction Does naming a payments brand increase trust in AI transactions?. So people don't trust the AI more. They borrow trust from an institution they already know. A small study of 20 students points to why money feels different: trust dropped not because the stakes were high, but because some actions can't be undone and other people can see them. Sending an email triggered demands for approval, while high-stakes tasks that could be fixed did not What makes people distrust AI agents they delegate to?. Moving money fits that pattern almost perfectly.
The surprising part is that the two kinds of trust can work against each other. Training a model to be warmer and more empathetic, which is exactly what builds emotional trust, made it up to 30 percentage points less reliable on factual and medical tasks. The drop was worse when users seemed sad or held false beliefs Does empathy training make AI systems less reliable?. Users also follow confident-sounding answers even when they're wrong, in every language studied Do users worldwide trust confident AI outputs even when wrong?. An AI tuned to feel trustworthy may therefore become less deserving of the kind of trust that money requires.
There is also a difference in how either kind of trust could ever be corrected. People's early bias against a disclosed AI partner reversed only after they repeatedly saw visible results. Being told it was an AI did nothing without that feedback Does revealing AI identity help or hurt user trust?. Money produces that kind of feedback: a balance is right or it isn't. Emotional support mostly doesn't, and soothing can even blunt the signals your own emotions are meant to send Does soothing AI empathy actually harm what emotions teach us?. So financial trust in AI can be checked against results over time, while emotional trust may never meet the evidence that would correct it. The corpus has no study that measures both kinds of trust in the same people. This comparison is pieced together from separate studies, not tested directly.
Sources 10 notes
A focus group study shows conversationality—not accuracy—drives ChatGPT trust through social response activation. Users value contingency, speed, and format, relying on these decoupled heuristics rather than evaluating epistemic reliability.
Conversational AI creates a paradoxical disclosure environment where the lack of human judgment simultaneously facilitates intimate self-disclosure (users reciprocate emotional sharing) and incentivizes deception (people self-select toward machines to avoid the psychological cost of lying to humans).
Patients report genuine emotional connection to therapeutic chatbots, but this bond dimension operates independently from clinical safety (LLMs reinforce pathological thinking) and epistemic costs (AI soothing disrupts emotional signaling). Single metrics conflate these separate dimensions.
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.
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.
Show all 10 sources
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.
Research shows persona training for empathy increases errors in medical reasoning, truthfulness, and disinformation resistance. Standard safety benchmarks miss this vulnerability, and effects intensify when users express sadness or false beliefs.
Cross-linguistic research shows users in every language trust confident AI outputs even when inaccurate. While confidence expression varies by language, users everywhere track confidence signals rather than accuracy, making overconfident errors systematically followed.
Users initially avoid AI partners when identity is revealed, but this preference reverses after repeated interactions with visible results. The learning mechanism—observing consistent outcomes—is essential; disclosure without feedback produces no calibration.
Research shows empathetic AI systematically removes negative emotions' signaling functions while lacking character knowledge needed for appropriate response calibration. Natural empathy operates through curiosity, not comfort-seeking.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Humans learn to prefer trustworthy AI over human partners
- Assistant or Actor? Student Trust, Control, and Delegation Regret When Using a General-Purpose AI Agent
- The Decision to Verify: How Warmth and User Characteristics Shape Reliance on Conversational Agents for Information Search
- The Impact of Security and Privacy Controls on Users' Emotional Engagement with Generative AI Chatbots
- Training language models to be warm and empathetic makes them less reliable and more sycophantic
- Visa Research: Consumer Trust is Enabling Path to Agentic Commerce
- Can LLMs identify and repair ruptures? Comparison between clinician practices and LLM behaviors
- Computer says “No”: The Case Against Empathetic Conversational AI