OpenAI's GPT-5.2 model cites Grokipedia | Hacker News Reader

OpenAI's GPT-5.2 model cites Grokipedia

engadget.com

49 pointsbhouston20 comments

Opus 4.5 often does this as well. I’ve been seeing it more days than not recently. It drives me nuts. Is it worse than citations that don’t exist or citations that are “real” but don’t actually contain any salient content? I’m not sure.

Don't worry, its "safety" features will shield all corporate liability while your kids are indoctrinated by 4chan written "history" regurgitated to you through Microsoft brand "AI"!

Shoe any example of such an article?

https://www.theguardian.com/technology/2025/nov/17/grokipedi...

https://www.nbcnews.com/tech/elon-musk/elon-musk-grokipedia-...

https://english.elpais.com/technology/2025-12-06/jimmy-wales...

My friend, your hero literally seig heiled on stage in front of the world. Do you think you are clever by lying to us? We can also read the white nationalist trash he reposts on "X". Why do you think he wanted a new Wikipedia in the first place. The cat is out of the bag, you are complicit or a fool.

I had duckduckgo return a grokapedia page for the first time. The search page has preview text making it seem like there was information so I clicked the link to check it out and it was a 404 page. What kind of SEO hack is that? Information for the crawler but nothing on the actual page?

It seems to be adding tons of articles, then some of them get deleted. I assume it's been allocated lots of compute. The entire model is outcompeting wikipedia on quantity per topic. If wikipedia merges/integrates some article and Grokipedia has a specific page for it, the search engine/LLM will get that version front and center. Grokipedia seems to have no scope limit, so wikipedia "non-notable" entries will be SEO-optimized towards sites with the topic-names, eventually settling on AI content farms as primary destination.

This is pretty dangerous because most people will not check the sources that LLMs are referring to. Grokipedia uses Grok and Grok trains on X, which is highly manipulated by both X itself (lots of people allege suppressed reach and other forms of shadow bans if you’re left of MAGA) and by bots. Not to mention the platform is naturally one sided when toxic content (racists, misogynists, outright supremacists, etc) drives away those of different views. For example look at Vivek Ramaswamy quitting X recently after all his posts got flooded with nearly 100% vile racist replies.

He deserves it unironically. I love his tweets and that he tried to speak truth to power to republicans (i.e. calling them/Americans out for being mean to nerds and not glorifying nerds like they do in asia), but he deserves to see that his brand of political thought almost always leads to extreme, virulent racism.

People who vote for leopards NEED to have their faces ate.

I don’t know. I felt he was the best version of what the right COULD be. Seeing that attacked with vile racism makes me sad.

Recently I asked an obscure question and it thought for awhile and it gave me a lot of output with sources.

Over half the citations were from Grok .. not even grokipedia .. just “share” pages from questions other people asked.

How can that work? They are not indexed by Search. They don't pop up in Google

Maybe someone is posting links to a webpage somewhere?

I don't have the exact question I asked it, but it was about the use of baby foreskin in beauty products.

I just asked it a similar and more simplified question: "are there beauty products that still use baby foreskin as an ingredient" .. and grok was 1/3 sources .. my initial question it was 3/6 sources.

Citations to pages like:

https://x.com/i/grok/share/2q3paAYZAsTD2OU5RAMxFEdid?utm_sou...

This sounds more like an issue with whatever web search/index tool GPT is using rather than the language model itself.

Musk is sama's hero, remember that.

https://www.courtlistener.com/docket/69013420/379/75/musk-v-...

This makes sense. I already use Grokipedia maybe 50% of the time. If you really dig into things, it is - incredibly - more accurate. I often find glaring errors or biases in Wikipedia, especially over the last 5 years.

Can you give an example?

are you sure that what you count as bias is actually just reality but you consider it bias.

And what's actually biased (see SilverElfin's comment) you consider fair

What is this psychology phenomenon called?

What’s wrong with grokipedia? I find it fair and reasonable. And it’s always nice to have a competitor because although Wikipedia is okay, it’s not sustainable to have it as a monopoly.

I like that OpenAI usds grokipedia and predict that Grokipedia will become more common and normal.

Ah yes Wikipedia, a non profit's a monopoly which generally has no bias and has become one of the best places even after the internet keeps on getting enshittified

Let's replace this with grokipedia whose grok generated some rather obscene and disgusting material on twitter as the main source instead of wikipedia.

...

Reply on news.ycombinator.com