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

2026-07-25T19:23:39Zc95ab4f60f143449c5c8cd8d8675a6c11b828c8acc8eafea8440f39619720d30
ai-securityandroidcertificate-transparencycookie-theftdbscdevice-bound-session-credentialsdns-parsergenaigoogle-workspaceindirect-prompt-injectioninfostealerllm-securitymemory-safetymerkle-tree-certificatespixel-basebandpost-quantum-cryptographypqcprompt-injectionrustsession-thefttlsvrpvulnerability-rewards-program

What happened

Collection of Google Security Blog posts (Apr 2026) covering multiple defensive and threat topics: 1) Indirect Prompt Injection (IPI) is a prioritized, actively-monitored AI threat; Google’s Threat Intelligence scanned the public web for IPI patterns and Google Workspace describes an ongoing, layered mitigation strategy for LLMs/agents. 2) Pixel modem security: Pixel 10 integrates a memory-safe Rust DNS parser into the baseband firmware to reduce memory-safety vulnerabilities in modem code. 3) Device Bound Session Credentials (DBSC) are entering public availability (Chrome 146 on Windows, mac/

Why it matters

A reviewed impact interpretation has not been published for this record.

Evidence and limitations

Source ID
google_security_blog
Record identifier
c95ab4f60f143449c5c8cd8d8675a6c11b828c8acc8eafea8440f39619720d30
Enrichment time
2026-07-25T19:23:39Z
AI-assisted enrichment
Yes

This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.