Forum Discussion
nhtkid
Nov 03, 2025Iron Contributor
Is Edge Copilot susceptible to Prompt Injection Attack?
I am getting lots of negative vibe when reading about OpenAI's Altas browser.
But what about Edge browser?
Let's say if you have Copilot in Edge turned on with Page Content allowed.
You asked the Copilot to summerise the page you are browsing and the page has hidden text to initiate a prompt injection attack.
Is it safe to say that as long as Copilot does not have control over your browsing or desktop, the damage is limited.
The worst thing it could happen is the change of AI behaviour, meaning the response returned from your question may not be what you asked for.
There is zero DLP risk, right?
No RepliesBe the first to reply