OpenAI says AI browsers may always be vulnerable to prompt injection attacks


Read on TechCrunch