A shield in your pocket
When it detects manipulative conversational patterns, the model will show an instant warning on the phone, urging users not to share sensitive information or transfer money. Because processing happens on the device itself, alerts arrive faster than with systems that must send data elsewhere for analysis.
Mark Chen, GovTech's director of trust and safety, said working with Google DeepMind brings advanced AI protection directly onto citizens' devices, catching highly targeted local scam attempts while keeping personal communications private.
“Fighting scams shouldn’t mean sacrificing personal privacy.” Mark Chen
Understanding local context
According to Google DeepMind authors JJ Geewax and Miguel de Andres, general-purpose AI models often struggle to identify scam tactics specific to Singapore. The new model has therefore been trained in a particular way.
















