79% vs 51%: independent neobanks lag on app security. Read the report

Security

Security

Mobile Banking App Security, 2025 vs 2026: What Changed

Banks fixed some of last year's most common app flaws, but biometric login got easier to bypass as it spread, and AI assistants brought new risks. What changed between our 2025 and 2026 banking app reports.

For most customers, the bank is now an app. They sign in with a fingerprint, approve payments and check their balance on their phone. Attackers follow them there: a weak check in a banking app is shipped to every customer who installs it.

A year ago, our Banking Report 2025 looked at more than 500 of the top banking apps on Android and iOS. The Mobile Banking App Security Report 2026 goes much wider: 2,686 apps from 487 banks in 64 countries, on Android, iOS and HarmonyOS. Set side by side, the two reports show where banks made progress, and where the risk has moved.

Last year's flaws: three less common, one more

Flaw 2025 2026 Change
Libraries with a known flaw 46% 38% 8 points better
Account takeover through sign-in links (OAuth) 14% 8% 6 points better
Unencrypted traffic allowed 20% 16% 4 points better
Biometric login can be bypassed 28% 30% 2 points worse

Lower is better on every row. Each year covers all the apps tested that year. They are not the same apps, and the 2026 sample is about five times larger, so read the changes as direction, not precise movement.

Libraries improved most: 38% of apps ship a third-party library with a known flaw, down from 46%. That is still more than one app in three, and in a typical Android app the libraries are a median of 3 releases behind the latest. OAuth flaws that can lead to account takeover fell from 14% to 8% of apps. The highest number of serious flaws found in a single app barely moved: 19 in 2025, 18 in 2026.

One 2025 headline is not in the table: hardcoded secrets, found in over half the apps in 2025. Our scanner's secrets rules changed this year, so the two figures cannot be set side by side. In 2026, 18% of Android apps ship a high-confidence secret key, such as a cloud or messaging service key, or a private key.

What banks added, and what is still missing

Banks have been busy hardening their apps. Biometric login now appears in 86% of apps, against over 65% in 2025, and a root or jailbreak check sits in the code of 70%, against over 40%. Fewer apps rest on very old foundations: 19% of iOS apps first shipped before 2014, against 47.7% in 2025.

The newer defences have not kept up:

  • Passkeys. Only 5% of apps offer passkeys, the sign-in method that resists phishing.
  • Tampering. 43% of Android apps notice a repackaged copy of themselves.
  • Scam calls. Only 11% of apps have every check against a scam call: they can tell a call is in progress, tell the screen is being shared or recorded, and show a scam warning.
  • New scanner checks. 59% of Android apps store sensitive data unprotected on the phone, 51% use weak encryption and 35% show web pages in an unsafe built-in browser.

Where the risk is moving

Three changes make this year different from the last.

  • Biometric login spread faster than it was secured. Bypassable biometric login went from 28% to 30% of apps while biometric login itself reached 86%. A fingerprint prompt that only returns yes or no can be skipped. The protected action needs to depend on a key that the biometric check unlocks.
  • AI assistants arrived. 3.7% of apps (100 of 2,686) now ship one. None of them leaked a key to an AI service, but 30% of them ship a secret key in their code, against 19.9% of apps without an assistant.
  • Updates slowed down. 85% of iOS apps and 86% of Android apps were updated in the last six months, down from 94.7% and 88.8%. The share of apps whose latest version is over a year old rose from 1% to 7%.

Many banks also share foundations: 14% of banks build their app on a shared ready-made platform, and the most common, Backbase, is used by 4%. For the first time, the report also covers banking apps from Huawei AppGallery, on HarmonyOS.

The gains of the past year point to steady work rather than new technology. Fewer flawed libraries and safer sign-in links are what regular patching and standard sign-in flows produce. The gaps that remain, passkeys, key-bound biometrics, tampering and scam-call checks, close the same way: one release at a time. Every release that ships without them leaves the same opening in every installed copy.

The 2026 report lists eight actions for banking app teams. Read the full report, or see the Banking Report 2025 for last year's figures.

How the two years compare

The 2025 figures are those published in the 2025 report. The 2026 figures cover all the Android and iOS apps scanned this year. The 2025 report does not list its apps, so the two samples cannot be matched app by app. A finding is a scanner rule match in the shipped package, not proof of exploitability. No app was attacked, and no bank or app is named.