Bank of America is a fixture in AI bank lists, from app features to fraud lawsuits

Measured on 2026-09-29 across ChatGPT, Claude, Gemini, Perplexity and Grok, Bank of America was named in 469 of 1,250 answers to 250 benchmark questions, none of which named a brand. In this snapshot only JPMorgan Chase was named more often among the sector peers. Grok named Bank of America in 145 of its 250 answers, against 72 for Claude, 75 for Perplexity, 87 for Gemini and 90 for ChatGPT.

Visibility varied by the intent of the question. It was highest on digital banking experience (59%) and leadership (57%), then safety (49%) and employer questions (45%). It was lowest on rewards and perks (22%), reliability (27%) and value (27%). In these answers, Bank of America is easier to find when the question is about apps, scale and workplace than when it is about perks or dependability.

The context split into three groups. On product and employer questions, the answers gave it credit. Perplexity singled out its in-app virtual assistant. ChatGPT pointed to its training organization, The Academy. Claude listed its single-use card numbers and next-day cash reimbursement among fraud protections. On fraud and conduct questions, the answers pulled in enforcement history. Claude and Gemini both tied it to the CFPB's Zelle lawsuit alongside JPMorgan Chase and Wells Fargo. Perplexity cited a $12 million CFPB penalty over discrimination against mortgage applicants receiving disability income. Grok named its 2014 settlement as the largest tied to the 2008 mortgage crisis. On service questions, Claude called out one gap: early direct deposit.

The tags for what Bank of America was credited for are a thin set next to its 469 appearances. The largest category was 'Not Substantive' (11), followed by Safety & Compliance (5) and Financial Performance (4). In many answers it is a name in a list rather than a brand with a distinct claim attached.

PeakMetrics' news index recorded 100,429 mentions of Bank of America over the last 30 days, and 11,436 of them had the brand in the headline. That is the second-largest news volume among the sector peers in this snapshot, close to Morgan Stanley's.

The sampled headlines are mostly market and corporate-calendar items. One reported a stock drop described as its worst since April 2025. Two were investor-filing stories about Bank of America Corp DE selling shares in other companies. One announced a company presenting at the Bank of America Financials CEO Conference. A fifth was a human-interest story reporting that a Bank of America vice president was named as a victim in a Times Square stabbing.

The press story is about markets, investor activity and events where the bank's name is incidental. The AI story is about consumers: which app is good, who protects against fraud, who was sued or fined, and who trains new hires. Little of the press volume in the sampled headlines matches the topics the assistants used when they named the bank.

Heavy coverage did not translate into a distinct reputation. Both the news and the AI answers place the bank in a crowd, but they are different crowds. The press crowd is investors and stock tickers, and the AI crowd is other big banks and issuers. In the fraud and mortgage answers, the assistants also drew on years-old enforcement material that the recent headlines do not touch.

This is one snapshot of live answers, so treat the figures as a baseline and not a ranking. Two things are worth tracking on repeat runs. The first is whether the Zelle fraud lawsuit and past settlements keep showing up on fraud and conduct questions. The second is whether the low-visibility topics, rewards and perks, reliability and value, stay thin. Those are questions where Bank of America is mentioned less often, so competing names fill the answer.

Cited domains and the specific claims attached to the brand matter more than raw mentions. A comms team would check which pages the assistants draw on for fraud protection, careers and app features, and whether those pages carry current information. A small gap like 'early direct deposit remains unavailable' is a single sentence in a single Claude answer, but on a service question it is the one thing said about the bank.