Tag
This paper introduces the first public multimodal dataset of 100 Turkish scam and benign phone calls, evaluating seven LLMs under raw audio, ASR transcripts, and human-corrected transcripts. Results show transcript-based inputs outperform direct audio, highlighting the need for inclusive AI safety research in low-resource languages.
Google is rolling out a new feature for its Phone app that detects and flags calls from scammers impersonating contacts using AI voice cloning, with automatic activation on Pixel phones running Android 12 and later.
ORACLE is a new agentic framework for early scam anticipation from streaming app usage trajectories. It uses a self-evolving context manager and on-policy self-distillation to detect scams from partial observations over multiple apps and days.
PreScam is a benchmark for modeling scam progression in multi-turn conversations, built from real-world scam reports. It includes tasks like real-time termination prediction and scammer action prediction, finding that supervised encoders outperform zero-shot LLMs.
OpenAI is launching a multi-year partnership with AARP and its Older Adults Technology Services (OATS) to help seniors safely use AI, including new educational content on spotting scams using ChatGPT and expanded training programs nationwide.