Back to blog

Segui e iscriviti

The AI Trust Gap: Why Consumers Trust Banks Over Fintechs When It Comes to AI

Low consumer trust in AI poses a risk for fintechs. Review the survey results to learn how Fastly AI Security prevents fraud without driving churn.

Alina Lehtinen-Vela
Alina Lehtinen-VelaResponsabile del marketing dei contenuti

Many consumers remain skeptical about whether digital-only financial providers can be trusted to use AI responsibly. Our recent survey found that consumers are four times more likely to trust traditional banks than digital-only financial apps when it comes to responsible AI use (24% vs. 6%).

Here are some more takeaways from our fintech consumer survey.

*With digital-only fintech companies, we refer to financial technology providers that operate exclusively through digital channels without physical branch locations.

Key Takeaways:

  • While 44% of consumers trust both equally, traditional banks are still favored 4x more than digital-only apps to use AI responsibly (24% vs. 6%)

  • 53% of consumers aren't confident their provider can tell them apart from an AI impersonator.

  • 66% are uncomfortable letting an AI agent move money without approving each transaction individually.

  • 72% of consumers would switch providers if their app falls behind in protecting against AI agent fraud

Why Consumers Trust Digital-Only Providers Less with AI 

Digital-only apps don't have the decades of brand history that traditional banks have. Traditional banks also tend to be more regulated. A 2024 joint notice from the OCC, Federal Reserve, and FDIC confirms that chartered banks fall under direct federal supervision, while many fintech companies aren't federally insured institutions. 

On top of this, AI adoption is moving faster than most consumers can evaluate it. Our survey data shows that only 6% of consumers trust a digital-only provider more than a traditional bank to use AI responsibly, compared to 24% who trust the traditional bank more.

This data lines up with broader consumer skepticism toward AI. Though AI adoption has climbed, trust in AI hasn't followed the same upward trendline. A 2025 University of Melbourne and KPMG study found that 66% of people now use AI regularly, but only 46% trust it. Americans tend to be even more skeptical. The 2025 Edelman Trust Barometer shows that only 32% of Americans trust AI, compared with 72% in China. Edelman's research also found that only 32% of Americans feel comfortable with businesses using AI at all.

The Agentic AI Fear is Already a Switching Trigger

Fastly’s research found that more than half of consumers — 53% — aren't confident their provider could tell them apart from an AI impersonating them. And 66% say they're uncomfortable letting an AI agent move their money without approving each transaction individually.

Agentic AI also leaves little room for error. 72% of consumers say they'd switch providers if their financial app falls behind competitors in protecting against AI agent fraud.

Banks are aware that detecting AI fraud is a key issue and is causing big challenges for the industry. Independent research from BioCatch found that 72% of banking leaders expect it to be very difficult to distinguish legitimate AI activity from malicious AI activity. 

How Fastly Can Help

The agentic AI fear respondents expressed in our survey demonstrates that consumers are worried about responsible use of AI, mistakes made by AI, and AI fraud. Investing in flexible, highly precise security solutions that also include features that help control and manage AI use is getting more important. Fintech companies need to control their shadow AI use so that it does not cause more security or cost issues. Fastly offers products that help teams manage and control their internal AI use and prevent AI attacks:

  • Fastly AI Firewall inspects every prompt and response passing through AI Runtime Control for prompt injection and isolates untrusted input so a malicious support ticket or a scraped web page can't quietly override an agent's instructions

  • Fastly API Security validates query parameters against the API contract you've already published. An agent, compromised or not, can't call an endpoint that isn't in your spec or pass a parameter outside its defined bounds; anything that doesn't match gets logged or blocked at the edge before it ever reaches your origin.

  • Fastly AI Runtime Control puts every model call behind a per-agent virtual key with its own rate limit and spend cap. An agent that starts looping, escalating calls, or spending outside its normal pattern hits its own ceiling instead of taking down anything else.

Ready to see how consumers view AI, fraud, and trust in financial services? Download the full Fintech Consumer Survey Report.

Pronto per iniziare?

Contattaci oggi