Google DeepMind and Singapore's GovTech have built an on-device scam detection model that runs on typical smartphones, fine-tuning a lightweight Gemma model on local scam conversations to catch region-specific deception. Scams account for over half of Singapore's reported crimes and more than S$4 billion in losses since 2019, with complex social engineering now making up 80% of losses. The model processes conversation data entirely on the device, avoiding server-side scanning that cannot inspect end-to-end encrypted messages, and triggers real-time warnings when it detects manipulative patterns. A two-stage pipeline uses a lightweight classifier for initial screening and the fine-tuned Gemma model for deeper contextual evaluation, balancing battery efficiency with detection accuracy.

Google DeepMind2 outlets

Teo told the same conference that GovTech is working with Google DeepMind on real-time detection of scams on people’s own devices.

Anthropic opened a new office in Singapore, its fifth in the Asia-Pacific region, and signed a Memorandum of Understanding with the Singapore Ministry of Digital Development and Information. The MOU outlines cooperation in AI safety, security, model evaluation, and responsible adoption across public and private sectors.

Anthropic2 outlets

OpenAI backs Singapore’s call for full transparency after an incident, but leaves it to the government to set how soon it must be told.

Computer Weekly