5 ms·

Just in case you were thinking of signing up directly with Moonshot to use the service, they appear to train even on API use: > We may use Content to provide,

by dovin 4mo ago

Just in case you were thinking of signing up directly with Moonshot to use the service, they appear to train even on API use: > We may use Content to provide, maintain, develop, support, and improve the Services, comply with applicable law, enforce our terms and policies, and keep the Services safe and secure. Customer who requires restrictions on the use of Customer Content for training or improving Moonshot AI models may contact Moonshot AI to discuss available enterprise arrangements or separate written agreements. Unless otherwise expressly agreed in writing, Customer Content may be used for the foregoing purposes. https://platform.kimi.ai/docs/agreement/modeluse#4-content https://platform.kimi.ai/docs/agreement/modeluse#4-content

onesandofgrain 4mo ago
You think openai, anthropic, google, z and any of the others dont? They do, if they say they dont, they do. Who wouldn't in this earth-shattering race. So Naive
lvillani 4mo ago
Interesting. OpenRouter classifies the Moonshot provider as ZDR. I wonder whether they have a ZDR agreement or it's a misclassification on their part.
- tigeroil 4mo ago
  My gut feeling is that Moonshot are probably ZDR but their terms are excessively permissive. That said, I wouldn't rule out OpenRouter misclassifying - I've seen some providers where I'm fairly sure they have.
- kzrdude 4mo ago
  OpenRouter's ToS also seems to allow them to store your submitted prompts anyway, so privacy advocates would have to look elsewhere anyway, that's at least how I understand it (and it surprised me).
  - ljlolel 4mo ago
    TrustedRouter (my site) has open source proof of confidential compute that we aren’t looking at prompts or output https://trustedrouter.com/ https://trustedrouter.com/
- kingo55 4mo ago
  Why risk it either way if they provide weights for others to run this? Am I being overly cautious not wanting to send my data to Chinese companies?
  - andrewinardeer 4mo ago
    Your safety is more at risk with your data in the US government's hands.
theplumber 4mo ago
[flagged]
- zyngaro 4mo ago
  The reason why the models are getting better is training on users conversations.
- nikcub 4mo ago
  > I pretty sure OpenAI and Anthropic are doing the same or worse. No they're not. It would end both companies if they were ever found to be doing that. Their terms are clear - if you use the coding plans they can[0] train in return. Enterprise and API, absolutely not. The argument here is that with the Chinese labs you have zero legal recourse. [0] opt-in, thanks
  - mahkeiro 4mo ago
    Are we talking about the company sending back private information through its client to « fight » model distillation?
    - theshrike79 4mo ago
      Yes. Enterprise contracts are checked and agreed by lawyers. The contract states no training. If the provider fucks up, there are actual monetary damages defined for breach of contract.
      - fluoridation 4mo ago
        It's an unenforceable clause. The affected party has no means to prove that a breach has happened.
        fzysingularity 4mo ago
        I think we all ought to look at the ZDR fine-print here. I get that in principle that there's no retention, but these are powerful models that can comprehend, paraphrase and summarize your logs for the sake of "product" improvement. Who knows what's collected here.
pietz 4mo ago
[flagged]
- bean469 4mo ago
  I would also assume the same for non-Chinese as well
  - TurdF3rguson 4mo ago
    The nightmare for Anthropic to be caught doing that combined with the temptation of their staff to virtue-signal by blowing the whistle... I trust them to act in their own interest if nothing else.
    - vb-8448 4mo ago
      And what if they use your data tò generate syntethic data to train on?
      - TurdF3rguson 4mo ago
        That would be just as bad.
    - knollimar 4mo ago
      Aren't they on the "we, the safe AI, must win at any cost" justifiers? What's a little contract violation if the fate of humanity is at stake?
  - villish 4mo ago
    Not for Enterprise. You can safely assume the trillion dollar companies would ban GPT/Claude from being used in house if that was a concern.
    - shunia_huang 4mo ago
      Or think it the other way, what if both A and O actually train with these data, then the enterprises found it(or maybe never), what's the other options? I'm not trusting these LLM companies because a. their model live with human generated data even if they claim to generate data with their own model, it's just nit human data, b. this is business not charity, eventually they live with customers' data, no exception
antiloper 4mo ago
This page says no, but the privacy policy is the authoritative document: https://www.kimi.com/help/kimi-api/api-data-security https://www.kimi.com/help/kimi-api/api-data-security
ouraf 4mo ago
If they want to take my awful prompts and poison their model with it, it's their loss. For anyone else that believes their input needs secrecy, you need to check the corporate plan of any provider for data protection clauses. Most use the cheaper plans as bait to get more training data and free feedback.

by dovin 4mo ago

onesandofgrain 4mo ago

You think openai, anthropic, google, z and any of the others dont? They do, if they say they dont, they do. Who wouldn't in this earth-shattering race. So Naive

lvillani 4mo ago

Interesting. OpenRouter classifies the Moonshot provider as ZDR. I wonder whether they have a ZDR agreement or it's a misclassification on their part.

tigeroil 4mo ago
My gut feeling is that Moonshot are probably ZDR but their terms are excessively permissive. That said, I wouldn't rule out OpenRouter misclassifying - I've seen some providers where I'm fairly sure they have.
kzrdude 4mo ago
OpenRouter's ToS also seems to allow them to store your submitted prompts anyway, so privacy advocates would have to look elsewhere anyway, that's at least how I understand it (and it surprised me).
- ljlolel 4mo ago
  TrustedRouter (my site) has open source proof of confidential compute that we aren’t looking at prompts or output https://trustedrouter.com/ https://trustedrouter.com/
kingo55 4mo ago
Why risk it either way if they provide weights for others to run this? Am I being overly cautious not wanting to send my data to Chinese companies?
- andrewinardeer 4mo ago
  Your safety is more at risk with your data in the US government's hands.

theplumber 4mo ago

[flagged]

zyngaro 4mo ago
The reason why the models are getting better is training on users conversations.
nikcub 4mo ago
> I pretty sure OpenAI and Anthropic are doing the same or worse. No they're not. It would end both companies if they were ever found to be doing that. Their terms are clear - if you use the coding plans they can[0] train in return. Enterprise and API, absolutely not. The argument here is that with the Chinese labs you have zero legal recourse. [0] opt-in, thanks
- mahkeiro 4mo ago
  Are we talking about the company sending back private information through its client to « fight » model distillation?
  - theshrike79 4mo ago
    Yes. Enterprise contracts are checked and agreed by lawyers. The contract states no training. If the provider fucks up, there are actual monetary damages defined for breach of contract.
    - fluoridation 4mo ago
      It's an unenforceable clause. The affected party has no means to prove that a breach has happened.
      - fzysingularity 4mo ago
        I think we all ought to look at the ZDR fine-print here. I get that in principle that there's no retention, but these are powerful models that can comprehend, paraphrase and summarize your logs for the sake of "product" improvement. Who knows what's collected here.

pietz 4mo ago

[flagged]

bean469 4mo ago
I would also assume the same for non-Chinese as well
- TurdF3rguson 4mo ago
  The nightmare for Anthropic to be caught doing that combined with the temptation of their staff to virtue-signal by blowing the whistle... I trust them to act in their own interest if nothing else.
  - vb-8448 4mo ago
    And what if they use your data tò generate syntethic data to train on?
    - TurdF3rguson 4mo ago
      That would be just as bad.
  - knollimar 4mo ago
    Aren't they on the "we, the safe AI, must win at any cost" justifiers? What's a little contract violation if the fate of humanity is at stake?
- villish 4mo ago
  Not for Enterprise. You can safely assume the trillion dollar companies would ban GPT/Claude from being used in house if that was a concern.
  - shunia_huang 4mo ago
    Or think it the other way, what if both A and O actually train with these data, then the enterprises found it(or maybe never), what's the other options? I'm not trusting these LLM companies because a. their model live with human generated data even if they claim to generate data with their own model, it's just nit human data, b. this is business not charity, eventually they live with customers' data, no exception

antiloper 4mo ago

This page says no, but the privacy policy is the authoritative document: https://www.kimi.com/help/kimi-api/api-data-security https://www.kimi.com/help/kimi-api/api-data-security

ouraf 4mo ago

If they want to take my awful prompts and poison their model with it, it's their loss. For anyone else that believes their input needs secrecy, you need to check the corporate plan of any provider for data protection clauses. Most use the cheaper plans as bait to get more training data and free feedback.