OpenAI Built GPT-Red To Attack Its Own Models Before Prompt Hackers Do
OpenAI has introduced GPT-Red, an internal automated red-teaming system built to find prompt injection weaknesses and make AI agents safer ...
Tag: prompt injection
Context Bombing Turns Prompt Injection Into A Defence Against AI Hackers
Security researchers are testing context bombs, defensive prompt-injection traps designed to stop malicious AI agents during cyberattacks.