AI threats in the wild: The current state of prompt injections on the web

Google security research highlights active and emerging threats across AI systems, browser session theft, cellular modem memory safety, and the transition to post-quantum cryptography. Key concerns include indirect prompt injection against AI agents and Workspace data sources, infostealer theft and replay of browser cookies, remote exploitation risks in modem firmware, and the long-term threat quantum computing poses to public-key cryptography. Google describes mitigations including continuous AI defenses, Device Bound Session Credentials, Rust-based modem components, and Android/Chrome post-量