Telcos are racing to put AI in the app. Their subscribers have already said which kind they will trust

Last week, the GSMA released its ASEAN Consumer Scam Report 2026 at M360 ASEAN in Kuala Lumpur. Nearly half of scam cases remain unresolved across six markets, and 90% of consumers recognize AI being used in scams. Julian Gorman, the GSMA’s head of Asia Pacific, emphasized that access alone is no longer enough. Just a month prior, the GSMA’s Mobile Economy Asia Pacific 2026 highlighted the rise of AI and declining digital trust as key forces shaping the region’s future.

Consumer trust in telecom operators has shifted from compliance to a critical churn metric. Scammed consumers are over twice as likely to switch providers compared to those who haven't been scammed. Thirty-five percent of consumers would switch financial services for stronger security based on telecom data.

Operators are now racing to integrate AI into their customer apps, using the same technology that erodes the trust needed to deploy it. Trust is now intricately tied to churn numbers. The 2026 scam report revealed that people who have been scammed are more likely to switch accounts or providers, while the 2025 edition found that 36% of consumers would switch financial services providers for enhanced security backed by telecom data.

Some operators, like CelcomDigi and Globe, are treating protection as a product. CelcomDigi sells a dedicated safety range, and Globe has formed an anti-scam coalition with the Global Anti-Scam Alliance. However, trust in AI within telecom apps is still a thin layer, resting on vendor contracts rather than mature regulatory frameworks.

Consumer comfort with data sharing in conversational AI apps fell from 53% to 40%, while comfort with limited, purpose-bound checks increased to 72%. These numbers represent the type of AI subscribers are willing to accept. Most consumers prefer background, purpose-driven AI using minimal personal data. Four tests can be applied to ensure trust before deployment: reading patterns, not people; holding the identity key; processing data in-country; and exposing AI as an operator-controlled API with logging, pricing, and revocable capabilities.

In designing AI solutions, breadth can substitute for personalization by using wide behavioral signals across telecom, payments, content, and gaming. Focusing on patterns rather than identity can personalize experiences without compromising privacy. Most subscribers evaluate trust through small, frequent interactions, such as credited points or functional vouchers. Each positive experience contributes to trust, while intrusive tracking can erode it.

Operators should ensure their security teams can test AI before launch and that it can be switched off at any time. If these conditions are not met, the AI is not ready, regardless of its performance. As Southeast Asia develops mature rules for networks and personal data, AI integration in telecom apps must prioritize transparency and control to build lasting trust.