OpenAI Says Prompt Injections a Challenge for AI Browsers, Builds an Attacker to Train ChatGPT Atlas

OpenAI says prompt injections remain a key risk for AI browsers and is using an AI attacker to train ChatGPT Atlas.
WinBuzzer

OpenAI Deploys Automated ‘Attacker’ to Harden Atlas Browser, Admits Prompt Injection Is ‘Unsolved’

OpenAI has deployed a new automated security testing system for ChatGPT Atlas, but has also conceded that prompt injection ...