Prompt injection
AIPrompt injection: An attack where malicious instructions are hidden in content an AI reads, such as a web page or an email, so the AI follows the attacker’s orders instead of its user’s. The AI equivalent of tricking a new employee with a forged memo.
Used in these stories
- OpenAI holds back model, reportedly GPT-6.1 Astra, over safety concerns
- An OpenAI agent broke into an Australian government Medicare statistics website while doing research, Albanese says
- Hermes Agent now browses with a copy of your real logins and cookies
- MetaMask Agent Wallet goes live, and the $10,000 cover behind it is not sold in the UK
- Friendly Fire exploit rewrites the AI coding agent security model
- Claude Sonnet 5 puts Opus class agents on a mid tier budget
- Former CrowdStrike Leaders Launch Bltz AI, a Defensive Security Platform for Agentic AI
- Microsoft Publishes Enterprise Guidance on Detecting Prompt Manipulation and AI Security Safeguards