Grok Bot now warns users about hidden prompt-injection instructions
It flags injected text on pages and won't follow it.

X Freeze (@XFreeze) said Grok Bot will now warn users about hidden instructions.
If a page or message tries to inject hidden text into the chat, the bot shows a warning and does not follow those instructions. Asked about a store's return policy, for example, it answers the real question and flags what it found.
Why it mattersPrompt injection is one of the main security risks for AI agents that read the web.
The post drew about 881 likes by Wednesday morning.




