The team's prompts contain sensitive proprietary design files.
My team loves Claude from @AnthropicAI . But this new policy of retaining prompts and usage is a red line...we simply can't give over our usage. Prompts contain our IP; literally all our design files and docs. Why would this ever have been ok? It's sad because everyone was looking forward to using the new model. Sigh.
Many users criticized Anthropic's prompt retention policy for eroding developer trust and risking IP leaks after building initial credibility, while some voiced support for the concerns and welcomed OpenAI's migration help.
@NaveenGRao @AnthropicAI let me know how we at openai can help!
https://developers.openai.com/codex/migrate
My team loves Claude from @AnthropicAI . But this new policy of retaining prompts and usage is a red line...we simply can't give over our usage. Prompts contain our IP; literally all our design files and docs. Why would this ever have been ok? It's sad because everyone was looking forward to using the new model. Sigh.
@NaveenGRao @martin_casado @AnthropicAI Anthropic has signaled all their future models of this level will have mandatory retention. It's not a stable position to hold. In my mind it's a race between market forces eliminating this vs regulation cementing it.
My team loves Claude from @AnthropicAI . But this new policy of retaining prompts and usage is a red line...we simply can't give over our usage. Prompts contain our IP; literally all our design files and docs. Why would this ever have been ok? It's sad because everyone was looking forward to using the new model. Sigh.
@mikeknoop @martin_casado @AnthropicAI If @OpenAI gives us similar capabilities without data retention…it’s obvious what to do. In my mind, potential IP leakage isn’t worth the illusion of safety. Clearer Anthropic is making a different calculation.
@NaveenGRao @martin_casado @AnthropicAI Anthropic has signaled all their future models of this level will have mandatory retention. It's not a stable position to hold. In my mind it's a race between market forces eliminating this vs regulation cementing it.
@NaveenGRao @AnthropicAI Sad indeed. You should try Factory Router so you can stay model independent and not need your whole team to switch tool every time a model company makes bad choices :)
@ejc3 @NaveenGRao @AnthropicAI They can view every saved convo. When you have certain things that you can't afford to leak, you can't just trust Anthropic to keep the data secure. Doubly so when they already revealed that they're okay silently sabotaging their own users
@RandianCapital @AnthropicAI I’m not afraid of Anthropic building our product. I’m afraid of them training a pattern they see from us to improve their models and that pattern appears when our competitors prompt the model to design something similar to us.
@NaveenGRao @AnthropicAI Can you talk about threat risk you are concerned with?
They explicitly say they aren't training on it, so how could the IP end up leaking?
@NaveenGRao @AnthropicAI "documents" as well.
and yet many reckless in law rushing in
@NaveenGRao @AnthropicAI GLM, MiniMax, Kimi…
@NaveenGRao @AnthropicAI How do you know they're not training off your data already?
I would assume anything you send to Anthropic will be used by them however they see fit.
@ejc3 @NaveenGRao @AnthropicAI The company that launches sabotage as a service is not a trustworthy company
@NaveenGRao @AnthropicAI Are they training on data from API and enterprise users now? I hadn't heard.
@NaveenGRao @AnthropicAI because that allows Anthropic to take your data and build a competing product.
@RyanKemper10 @NaveenGRao @AnthropicAI Their business would be literally over if they violate their terms of service.
And your data is likely not that important to risk it.
@GlenWilsonIA @AnthropicAI We consumed all our models through @databricks, not first party serving. I know how that’s built…
@RobotChocobo @NaveenGRao @AnthropicAI Sue a trillion dollar company as a service will keep them in check.
@NaveenGRao @AnthropicAI they have a history of suspicious timing, including their Chief Product Officer remaining on the Board of Figma until days before they released a competing product. Not sure why companies are so trusting of this company.
@NaveenGRao @AnthropicAI at least they didn't insult your intelligence by calling their new marketing strategy "safety"...you didn't get the bird they flipped us biomed people
@Drachs1978 @AnthropicAI If there’s an incentive and it’s technically feasible, nature finds a way lol
We consume the models through databricks to avoid some of this issue
@RandianCapital @AnthropicAI I don’t trust any company :)