Their whole point is that they know that delivering this message (that the Google AI says only bad things about Firefox) is the intention of the image, but they don’t believe that this is what the AI actually does for a neutral prompt without seeing the prompt themselves.
Yeah, I’m not giving any credence to that without seeing what biases you put into the prompt
You don’t necessarily have to put any biases into the prompt, if the AI overview in question comes from Google or Microsoft.
Their whole point is that they know that delivering this message (that the Google AI says only bad things about Firefox) is the intention of the image, but they don’t believe that this is what the AI actually does for a neutral prompt without seeing the prompt themselves.