May 20 news, Apple's latest 2025 App Store Compliance Review Report shows that, under the collaborative governance of AI technology and manual review, the platform successfully prevented over $2.2 billion in potential fraudulent transactions in 2025, with a cumulative total of more than $11.2 billion in fraud intercepted over the past six years. As generative AI has significantly lowered the barriers to app development, malicious software and fraud are showing trends of large-scale and complexity, prompting Apple to continuously increase its R&D investment in machine learning tools.

iTunes, appstore, Apple App Store

In the past year, Apple used AI models to accurately identify complex malicious patterns and analyze app variants, rejecting 22,000 and 443,000 app submissions for fraud and privacy violations respectively, and removing nearly 59,000 apps suspected of "bait marketing" violations; its security defenses also extended to the periphery of the ecosystem, intercepting 28,000 illegal apps from pirated stores, and preventing 2.5 million high-risk submissions through TestFlight.

In terms of account management, the AI mechanism helped intercept 1.1 billion fraudulent customer accounts, deactivating 40.4 million abusive accounts and 193,000 developer accounts suspected of fraud.

Although the combination of AI and human efforts has greatly improved the efficiency of ecological purification, the platform still faces severe challenges. In April this year, a counterfeit cryptocurrency wallet app bypassed the review process, causing users about $9.5 million in losses; at the same time, AI-generated synthetic content and "nude" apps continue to test the limits of the review. This series of confrontations indicates that, in the face of AI-driven automated cybercrime, the governance of application ecosystems by tech giants has evolved into a long-term technological arms race. How to build a trust barrier in the automated attacks and defenses remains the core issue facing the entire mobile application industry.