GPT-Red hits 84% in red-team test
OpenAI's GPT-Red is an internal automated red teamer that attacks prompt-injection weaknesses and feeds those attacks back into GPT-5.6 training.
Prompt Injection News and Analysis
3 verified stories covering Prompt Injection, product updates and industry developments.
ChatGPT Lockdown Mode cuts off data exits
The new security setting does not solve prompt injection, but disables browsing, agents, connectors and other outbound channels that attackers need to exfiltrate data.
Opus 4.5 Introduces New Approach to Prompt Injection Defense
Anthropic's Opus 4.5 model improves prompt injection resistance through system prompt prioritization, context isolation, and refined refusal strategies, though no model is fully immune.