OpenAI’s Outlook on AI Browser Security Is Bleak, but Maybe a Little More AI Can Fix It

The company behind ChatGPT says prompt injection vulnerabilities will likely never be fully solved.'