Bank reputation holds at 70.5 while the measures that pay the bills slip faster

RepTrak's 2026 banking reputation score landed at 70.5 on a one-to-100 scale, down 0.4 points from 70.9 in 2025, keeping the industry inside the firm's "strong" range ^[1][2]. The number worth reading is underneath it: the business outcomes that describe what customers actually do fell further than the headline did, according to RepTrak senior director Sven Klingemann ^[3].

Those outcomes are willingness to do business with a bank, giving it the benefit of the doubt, and trusting it to fix a product or service problem ^[3]. Klingemann said the declines there were steeper than the reputation decline itself ^[3]. The survey was fielded in May and collected 14,000 ratings from members of the informed general public who were familiar with the banks they rated ^[4]. So this is not drift among the disengaged. It is erosion of crisis-time credit among people who know the institutions.

The deregulatory backdrop is doing some of that work. Only about 35% of respondents completely trusted banks to self-regulate in a deregulatory environment, which leaves roughly two-thirds withholding complete trust ^[5][6]. Around six in 10 said they were very concerned about changes to banking regulations ^[7].

AI is doing the rest. Just 32% completely trusted banks to use AI ethically if regulatory guardrails were reduced, with large banks performing particularly poorly, and 56% said they were very concerned about easing AI regulations at banks ^[8][9]. About a third of respondents said they were very likely to use AI tools for banking or personal finance, so this is not refusal so much as use without confidence ^[10]. On net agreement scores, the strongest sentiment was that AI needs more regulation and oversight at 53 points, followed by job losses at 39 and digital privacy at 35 ^[11][12][13]. AI improving customer service scored 4 ^[14]. That is a 49-point spread between what the public fears and what it credits banks for building ^[15]. Klingemann said the demand for more AI regulation was the number one concern across Republicans, Democrats, men and women ^[16].

The asymmetry is priced. Respondents aware of a bank failing to implement AI ethically would be expected to cut its reputation score by 8.5 points, and a failure to prevent or respond to AI-enabled fraud, including scams and deepfakes, carried an 8.4-point hit ^[17][18]. One visible ethics failure is worth roughly 21 times this year's entire industry decline ^[19].

RepTrak also asked ChatGPT and Google Gemini to score seven banks, and found machine-generated reputation scores far below the public's for every bank tested; TD Bank's public score was 51.3 points higher than the chatbots' ^[20][21]. Across the seven, the models averaged 42 on products and services against the public's 71, and 26.1 on conduct against 68 ^[22]. That is a 41.9-point gap on conduct ^[23]. ChatGPT was generally more negative than Gemini, and leaned more on government, regulatory and public-agency sources while Gemini leaned on news and editorial content ^[24][25].

What to watch: whether the soft measures keep leading the headline down in the 2027 print, because willingness to do business is the number that shows up in acquisition costs before it shows up in a reputation index. Watch also whether any bank manages to move the customer-service net score off 4 ^[14], and how the source mix in machine assessments treats an enforcement action in a lighter-touch supervisory era ^[24][25].