AI is becoming both a threat and a tool in smartphone security. As phishing attacks grow more sophisticated, vendors are increasing their focus on on-device AI protections. This blog explores what that means for the future of smartphone security. AI-generated phishing attacks pose a big threat to consumers and businesses if device vendors don’t fight back with more sophisticated on-device natural language checks and deepfake detectors. According to the latest findings in Omdia’s Mobile Device Security Scorecard, phishing attacks continue to be the most common smartphone security threat to consumers. Yet, hands-on testing of the latest devices, including the iPhone 17 Pro Max, showed that no device could detect and stop the kinds of sophisticated, hand-made phishing attempts which AI will make more common. 27% of consumers reported experiencing phishing scams in the past year - according to the Omdia Mobile Device Security Survey conducted in October 2025 - making it the most common type of security incident on smartphones. The survey also showed that phishing attacks are most common in English-speaking countries, with the United States leading with 40% of respondents experiencing phishing attacks, followed by the United Kingdom (36%), Ireland (35%), Canada (32%), and Australia (30%). Singapore, where English is one of the official languages, also reported a high rate of 26%. This trend suggests that scammers prioritise targeting English-speakers, likely due to the language's global prevalence, with a particular focus on high-income consumers in mature markets. The threat that AI poses to smartphone security As more advanced AI and larger LLM models become more accessible and affordable, scammers will have more tools in their belt. They can craft more sophisticated scam campaigns quicker than ever before, and can even tailor it to each target with a greater understanding of more believable language and formatting - within Omdia’s