OpenAI partners with major news publisher Axel Springer

Community
,
21

And why not?

It is reporting on an entertainment product as if it was a real thing.

“a man swung a stick at a ball today, which has been done heroically millions of times without consequence to the world”

1 Like
22

Right? The real news would be the man biting the dog, not a dog biting a man! :wink:

23

I would like to know if there are any details on this anywhere.

2 questions:

Does this apply to the API?
Is there opt out?

Thank you.

media brands like POLITICO, BUSINESS INSIDER, BILD, and WELT, including paid content.

What in the actual.

24

Greetings from Germany,

Don’t trust these newspapers!

1000038531
10000385311080×1373 208 KB

4 Likes
25

The API isn’t burdened with internet access you don’t provide yourself.

2 Likes
26

What the actual F***? I was chilled with all the Sam Altman hiccups, the perceived degradation in quality and length of GPT-4-Turbo did not bother me, but this really makes me doubt the direction OAI is heading.
Wow this is really bad news. Its Germany’s largest, most populist, right-leaning newspaper tightly integrated with the best LLM application of the world. The best possible outcome is that there will be 5 more such partnerships from other giant media outlets with opposing biases soon. But in my eyes curating knowledge sources should be none of OAIs business.

6 Likes
27

Agreed. It makes me wonder more about the ousting that occurred, that apparently nobody still knows why it happened. Then we receive this kind of news. These kind of negotiations takes quite some time to come to agreement.

I’d be pretty frustrated learning about these types of negotiations.

This is a stepping stone towards a terrible direction.

4 Likes
28

I’m not as worried. to get somewhere you’ll have to put one leg in front of the other, I don’t really care which leg I’m starting with, left or right, I care about the destination.

No spoilers, but I think this is the case, I’m aware of multiple different news entities that have expressed an interest in this :laughing:

1 Like
29

I don’t see a problem with someone being biased. Everyone knows that everyone is biased, and most biased are those who claim to be unbiased, thus making themselves biased… welcome to the real world.

What concerns me more is the possibility that Axel can now claim some copyright over any answer if he has ever written on a similar topic because there will surely be at least a few words marked as sourced from somewhere, and if we publish something ourselves, it will have to be visible, which eliminates the possibility of using ChatGPT for creative purposes. I think this is a big step back in terms of creativity, and a big step forward in starting to charge for something in the end, neither OpenAI nor Axel entered this project to lose out in the long run.

30

Isn’t this like when you ask Bing Chat for current events and it will give you an output that are collated from some news sources. And it will list the news sources at the bottom: Learn more:…

1 Like
31

I generally agree. That they started with this ugly ingrown toenail of the right food called the “BILD” may have sparked my rage a little. But my main concern is the destination: OAI curating information.

Except now you need a partnership and not just some SEO plus relevant content to be featured. Partnerships will be limited (otherwise it won’t make sense to have a partnership) and - without knowing the details - I assume that only a handful large publisher can afford to give away content/ are interesting for OAI to partner with.

That’s why I think it’s necessary to incorporate as many voices as possible into a tool. The model started with being trained on “the internet” which is approximately as many voices as you can get. Not saying the base model is unbiased but at least I know it reproduces the biases that are incorporated into (internet) society, not some spin on a current event.
Now they stack a layer on top, maybe 1 voice, maybe 20, but def. a limited number, which will become the source of truth for many people.

I would prefered to see them focus on providing excellent models and convenient ways to build on to of them instead of this.

32

Well, it cannot be helped that they need to tie up with media companies officially since some news companies disallow OpenAI to use their content or crawl their site.

1 Like
33

At a glance, this seems to be related to the data partnership program they announced in November.

Is anyone aware of any other partnerships they have already announced?

1 Like
34

Let’s try to answer some of the most frequent questions that we have gathered so far:

  1. Will there be other publishers partnering with OpenAI?
  • The deal is not exclusive, other publishers can enter a similar agreement.
  • Apparently News Corp is currently in advanced discussions to reach an agreement how their content can be used for generative AI.
  • There is already another deal with Associated Press (AP), which had been struck in July. But, it differs as AP is licensing parts of their archives for model training. In turn AP will get “access to OpenAI’s technology and product expertise as part of the deal”. It is not about content in the same way that the deal with Axel Springer is. But the deal with Axel Springer also includes the option for OpenAI to license their content for model training.
  1. How will this be implemented?
    We can only really know when this is being rolled out in the first quarter of 2024. As of today we can get this information from media reports:
  • ChatGPT will have access to “selected global news content” for summaries and also provide a link to the source. The linked source may actually be behind a paywall.
  • News summaries will be updated in real time. Users could replace other parts of their internet usage with going to ChatGPT and asking about the most recent events.

We learn that

  • one of the intentions is to drive traffic to a partner. With all the attached implications of including content in this way being similar to sponsored content.
  • this is integrated with a solution to pay publishers for providing content for model training.
  • this could be a way forward to integrate ads into regular ChatGPT conversations

Sources:

35

There’s a lot of noise about dropbox.

The whole pattern is pretty disturbing.

1 Like
36

I completely disagree with the logic of “others will make it better”. Of course, I could be wrong but typically it goes like this.

  1. First platform is given a “favorable position” in results.

  2. Next platforms join and demand fair share.

How do they accomplish a fair share? Tracking & targeting. Just look at history. Under the guise of “to serve you better content”, “aligned with your interests”. This is guaranteed to be the next step once enough platforms have signed up if this is the route to be taken.

Then, as with any typical tech company, we begin phase 2 of the enshittification

I’m not saying this is inevitable. I just ask people to keep this in mind and be cautious

2 Likes
37

I’m German.

They are literally the Fox News of Europe. Very disappointed. They spread so much fake news with BILD in Germany, disrespect dead people, hell they even annoy the parents of dead people!! I am considering to cancel my Plus subscription because it WILL affect the quality of Browsing in TONS of ways. GPT-4 already got lazy as hell, and with this step, you are just digging your own grave even deeper.

I do not want to support a company that decides TO STALK THE BEREAVED OF MULTIPLE MURDER CASES, actively DECIDES TO SPREAD FALSE INFORMATION ABOUT LGBT PEOPLE (talking about the “scientist” that witch hunts trans people), AND so much MORE.

4 Likes