A new study by IMDEA Networks (PDF) provides the evidence. The included infografic is from blog post by one of the study authors. Important details from his blog post:
It makes little difference if you do not accept cookies.
It does not matter whether you pay or not.
It does not matter if you use Brave and ad blockers.
I couldn’t find any german source on this topic yet.
- Datadog is an infrastructure monitoring tool
- intercom is user feedback/sales funnel service
- sentry is an exception/error capture service
- firebase is Google’s app development platform
- adjust is analytics/marketing platform
None of those are used to sell 3rd party ads and all of them keep the data for the companies who pay for their services (in this case ai companies).
I didn’t expect to defend ai companies, but this graphic is just stupid
The graphics does not claim that Datadog, intercom, sentry, firebase, adjust are selling ads. The graphics says that these companies use web trackers (see label in the right top corner).
The post title is my conclusion from the graphics. More precisely, all (listed) chat bots sell to at least one advertising company. This is clearly supported by the graphics. My claim was not that all chat bots sell to all advertising companies or that all web tracking companies are advertising companies. I hope this helps.
an FYI: ublock origin has an AI filter option (not enabled by default):
This chart would really benefit from color coded lines instead of just grey and magenta
It is color-coded?
Good to know! I was just about to checks notes give them my intimate data
Can you make the chart more ugly than it is now?
I tried but painting over all those terrible words didn’t make it worse
Picasso be like.
Are these white hands??
If you can’t trust your nonfungible ape girlfriend, who can you trust?
Not my intimate data!
:0
Never ever speak to an AI in first person. Never, “what do you recommend I put on my pasta tonight?”. Instead, “what are some good pasta toppings”. Most AIs will try to trick you and pull back to assuming personalization. Occasional reminders to never assume the conversation is about you do help minimize this.
Your data will still be harvested as if your own, but it introduces plausible deniability and minimizes linkages to build personal profiling.
Is this some proven tactic? I can’t imagine they would ever care if you use first person or not. If you start googling for Marmite you will he branded as a Marmite eater, regardless if you buy it for yourself or a friend.
The AI does create a personal profile of you and tries to categorize likes, dislikes etc. Hard to say what happens on the back end, but you will get less linking / leading prompts with depersonalized queries.
So this claims full chat sessions get sent to all of these services? I doubt it…
If i had to guess. They have another ai that analyze your logs and make a profile on you. Then they share certain things like interests etc with those who buy them. Probably packages
That’s quite a claim. I’d love to see some proof that they have an ai model profiling you and sending that profile to datadog.
Unless we get Snowden 2 i doubt it
Some AI chatbots send full chat sessions to some of the listed services. Others send just summaries of the important (incl. intimate) data.
Fork found in kitchen
But also this graph is also misleading, as some of these are services unrelated to advertising.
the information becomes relevant for advertisers real quick after it’s been sold to the right company
