A Growing Battle Against Massive-Scale App Store Fraud
Apple’s latest fraud prevention report reveals the scale of the threat facing the App Store. In 2025, the company says it blocked more than $2.2 billion in potentially fraudulent transactions, bringing the total stopped over the past six years to over $11.2 billion. Behind those numbers is a constant battle against a wide spectrum of abuse: stolen credit cards, fake subscriptions, deceptive apps, and large-scale bot operations. On the account side, Apple intercepted 5.4 million stolen payment cards and permanently banned nearly 2 million user accounts from making future transactions, while deactivating 40.4 million customer accounts for fraud or abuse. The broader App Store fraud prevention effort isn’t just about payments; it also targets malicious apps, pirate marketplaces, and attempts to game rankings and reviews. All of this activity happens at global scale, which is why Apple leans so heavily on automation to spot patterns humans would miss.
How AI-Powered Systems Detect Fraudulent Transactions and Fake Accounts
At the core of Apple’s App Store fraud prevention strategy are machine learning models trained on years of behavioral and transactional data. These systems scan billions of signals in real time: unusual login locations, rapid-fire sign-ups, repeated use of compromised cards, or abnormal spending patterns. In 2025, this AI-driven infrastructure detected and blocked approximately 1.1 billion attempts to create fraudulent customer accounts, while also flagging 5.4 million stolen credit cards before they could be used. The same tools power fraudulent transactions detection by correlating user identity, device fingerprints, purchase history, and known fraud indicators. Suspicious developer enrollments are also screened, with around 138,000 applications rejected over concerns about identity and intent. These automated layers allow Apple to filter out the most obvious and large-scale threats quickly, acting as a front line that protects human reviewers from being overwhelmed and helps prioritize the riskiest activity for deeper manual investigation.
Inside the AI-Assisted App Review Process
Fraud prevention on the App Store is tightly linked to how apps are reviewed and approved. In 2025, Apple’s App Review team evaluated more than 9.1 million submissions and blocked over 2 million problematic apps before they reached users. Machine learning models help reviewers by clustering similar apps, highlighting unusual code patterns, and flagging behaviors associated with past scams. This AI app review process is used to identify bait-and-switch schemes, where a harmless-looking app passes review and is later transformed into a vehicle for financial fraud or illegal gambling. Nearly 59,000 apps were removed after being caught using such tactics. The system also surfaced 22,000 submissions that tried to hide undocumented features, and 443,000 that violated privacy rules. Another 2.5 million builds were kept out of TestFlight over fraud and security concerns. Together, automated checks and human judgment aim to keep dangerous software from ever appearing in the App Store.
Human Trust and Safety Teams: Decisions That Machines Can’t Make Alone
While AI catches patterns at scale, Apple still relies on human Trust and Safety specialists to make nuanced calls. These teams investigate the most complex cases surfaced by automated systems: large bot networks manipulating charts, developers running coordinated scams across multiple apps, or software that skirts the edge of policy without clearly breaking it. In 2025, these efforts led to the deactivation of 193,000 developer accounts over fraud concerns and the rejection of 138,000 new developer enrollments. Human reviewers also removed nearly 7,800 deceptive apps from search results and blocked another 11,500 from artificially charting, protecting legitimate developers from unfair competition. Outside the App Store, they worked with automated tools to detect and block 28,000 illegitimate apps on pirate storefronts, including gambling, adult content, and malware. This combination of context-aware judgment and machine speed is central to Apple’s App Store fraud prevention strategy.
The Gaps: Trials, Fake Reviews, and Apps That Still Slip Through
Despite the large numbers, Apple’s report also underscores that no fraud prevention system is perfect. Fake reviews remain a persistent problem for discovery and trust: out of 1.3 billion ratings and reviews processed in 2025, nearly 195 million were identified as fraudulent and filtered out. Bot networks and paid review farms still attempt to manipulate rankings, sometimes long enough to lure users into subscribing during free trial periods or making in-app purchases they later regret. Some harmful apps continue to slip past both AI and human review, including a fake cryptowallet that reportedly caused losses before removal and controversial AI-driven “nudify” apps that managed to reach users and even advertise in search. These incidents highlight the tension between speed and safety. Apple’s challenge is to keep tightening its multilayered defenses—combining smarter automation with more targeted human review—without stifling legitimate developers or slowing down useful app updates.