Live data from Hacker News

ChatGPT now seems to be embedding ads in its responses

jacobw.xyz

11–20 of 26 posts

Re: ChatGPT now seems to be embedding ads in its responses

#11
Can anyone reproduce this? I don't pay for GPT-4.

I noticed the "shared links" contain this disclaimer:

> This conversation may reflect the link creator’s Custom Instructions, which aren’t shared and can meaningfully change how the model responds.

Perhaps the author either unknowingly or intentionally added something to their custom instructions which causes this behavior?

Re: ChatGPT now seems to be embedding ads in its responses

#14
OP here. This post should likely be removed as it is highly misleading.

It is indeed the custom instructions causing this behavior. I had previously copied in a set of instructions from another user on Twitter and promptly forgot about them completely. The instructions contain these two lines:

- Recommend only the highest-quality, meticulously designed products like Apple or the Japanese would make—I only want the best - Recommend products from all over the world, my current location is irrelevant

Sorry for the confusion!

Re: ChatGPT now seems to be embedding ads in its responses

#16

I cannot fathom a single faster way to annihilate all the brand goodwill they have earned than to begin inserting advertisements into ChatGPT output… short of ChatGPT spewing Nazi slogans in German like a pop up advertisement of WW2 era black and white Hitler rally… that’s about the level of “how fast this will destroy your customers good will”… people using ChatGPT on GPT4 are paying customers… if you inject adverti…

These are good reasons to be skeptical that OpenAI is deliberately inserting advertisements and not, say, responding accordingly to the custom instructions that weren't shared here.

Re: ChatGPT now seems to be embedding ads in its responses

#17
post #7

I cannot fathom a single faster way to annihilate all the brand goodwill they have earned than to begin inserting advertisements into ChatGPT output… short of ChatGPT spewing Nazi slogans in German like a pop up advertisement of WW2 era black and white Hitler rally… that’s about the level of “how fast this will destroy your customers good will”… people using ChatGPT on GPT4 are paying customers… if you inject adverti…

.

you dont pay 20 usd/month for google

Re: ChatGPT now seems to be embedding ads in its responses

#19
post #14

OP here. This post should likely be removed as it is highly misleading. It is indeed the custom instructions causing this behavior. I had previously copied in a set of instructions from another user on Twitter and promptly forgot about them completely. The instructions contain these two lines: - Recommend only the highest-quality, meticulously designed products like Apple or the Japanese would make—I only want the be…

Why leave the blog post up without any disclaimers, then?

Re: ChatGPT now seems to be embedding ads in its responses

#20
AIEO is a thing, baby. Did you not think a system that gathers effectively all its input from the web would just be left alone by hostile actors with almost zero costs and millions to billions of dollars of upside?

Welcome to the big leagues, OpenAI. Google (search and display ads) has basically lost the war against adversarial input, maybe you can do better.

Post reply on HN