OpenAI Warns Prompt Injection Threat Persists
By
Marcus Lee
OpenAI acknowledges prompt injection attacks cannot be fully eliminated and is using automated AI attackers to harden its Atlas agentic browser against threats.