DeepSeek found to be sharing user data with TikTok parent company ByteDance

Gee I wonder if grok shares data with the rest of twitter, Gemini shares with the rest of google or llama… actually I haven’t used Facebook in ages, I don’t even know if there’s a ChatGPT equivalent service on Facebook.

I do actually wonder if Anthropic shares data with Amazon, or OpenAI with Microsoft (their majority shareholders). That would be a direct 1:1 comparison with what’s happening between deepseek and bytedance (though at least in the latter case you can host-your-own since the model is open source).

DeepSeek isn't owned by bytedance, which is why it's news worthy. The title is stating DeepSeek (an independent company) is sharing data with Bytedance (TikTok parent company).
So it would be like OpenAI sharing ChatGPT data to Google. Which would be noteworthy as there is likely a lot of sensitive information that people share and most people probably don't think that data is being sold/shared.
- Yes, that’s a valid distinction. Though practically speaking, I don’t really know how different it is from anthropic sharing with Amazon, for example
Chinaman bad bro!
Llama is touted as better than chat gpt. It ostensibly doesn't share data if you self host it... But I never trust any of them.
- Llama is not touted as better than ChatGPT models. On most benchmarks Llama3.3 performs worse than ChatGPT 4o and much worse than OpenAi's reasoning model o3-mini.
  Unless you're speaking in terms of privacy and then unfortunately no one really knows as OpenAI does not release any of their metadata and Llama models are self hosted so there's no method in which the data will leave the LAN they're hosted on. It's a case of apples and oranges.

This is the web chat client/app, just like OpenAI sharing data with Microsoft, or Copilot doing the same. If you self host these LLMs your data stays within your LAN.

I'm moving to self host all my streaming stuff. Switching from local-only plex to self hosting all my media (spotify, google photos, LLMs) and tools behind a reverse proxy so i can access outside my home. It's pretty sweet and a good learning experience using reverse proxies
Edit: Plus fuck these technofeudal lords who enclose access to markets, information, and culture.
- !selfhosted@lemmy.world to the win!
- Love to see it.
  Edit: replied to the wrong comment
- A freedom enjoyer detected. A Proper authority has been notified.
You can't practically self-host Deepseek R1.
Look, I use the 32B distil on my 3090 every day, but it is not the same thing as full R1. And people need to stop conflating the two.
And (theoretically) API usage through one of many R1 providers is private.
- Do you know of a provider is actually private? The few privacy policies I checked all had something like "We might keep some of your data for some time for anti-abuse or other reasons"...
- I use 32b and the 672b side by side. The performance hit is around 20% and I keep all my data local. I am not conflating the two however self hosting works for me just fine. Your usecase is your own certainly. However I'd rather take the performance hit for the added data privacy.
  Also it's nice to he able to set my own weights and further distil R1
  I have a local python expert a local golang expert and both have my local gitlab repository and I've tied their respective Ollama keys to my VSCode IDE.
- I dont really use LLMs so I didn't even realize there were versions with different weights and stuff. I was using 7b, but found it pretty useless. Pretty sure I'm not going to be able run 32B on my rig. lmao.
  guess ill continue being an LLMless pleb.

what's the least surprising thing you've heard all day

This is the worst AI, just in case anyone is wondering. The only thing that's interesting is that you can see it reasoning and thinking before it answers questions.

Otherwise, it's just trash. It makes huge mistakes, it hallucinates often, and it takes more processing power to do those things than other popular alternative AI's.

All LLMs lie, this is why it's important to verify what output you're getting. A GPT is essentially text prediction that has been trained on a very large dataset, think of when you end up sending "ducking autocorrect" in a text. Furthermore, Deepseek has distillations of many models. Which do you have experience using?
https://github.com/deepseek-ai/DeepSeek-R1
Edit: To add even more context GPT and Diffusion models are patently not AI as they are not able to verify what output they're giving. These are all tokens that are fed into a recursive algorithm. They're vector database queries that have reinforced pathing. None of these "A.I." models are thinking or reasoning, yet.