OpenAI accusing DeepSeek of intellectual property theft for allegedly training on GPT output is a classic case of “the pot calling the kettle black,” especially given the lawsuits against OpenAI for using copyrighted material in its own training data. - ThreadSky

carnage4life.bsky.social • 31 days ago

OpenAI accusing DeepSeek of intellectual property theft for allegedly training on GPT output is a classic case of “the pot calling the kettle black,” especially given the lawsuits against OpenAI for using copyrighted material in its own training data.

1 / 2

Comments

bartlebyscorpse.bsky.social•31 days ago

Appropos of nothing, it also seems like another sign of encroaching AI enshittiffication, as AI models simply train on the responses from AI models in a Circular Human Centipede of reprocessed output

ggsake.com•31 days ago

Evergreen

gregbrailsford.com•31 days ago

I one of those who strongly believes training AI on copyrighted works is completely protected by fair use (provided the works are legally purchased) so it's funny that OpenAI thinks they can train on existing work but DeepSeek can't?

magfilm.bsky.social•31 days ago

We stole it first so it's ours!

mingo.nl•31 days ago

Isn’t that Bill Gates’ line?

zomby2d.bsky.social•31 days ago

I believe it was Steve Jobs

baxate.bsky.social•31 days ago

Wasn’t there a decision from the Copyright Office that LLM output can’t be copyrighted?

danjac.bsky.social•31 days ago

Libertarians again going crying to Daddy Government, see also the SVB bailout.

clorabell.bsky.social•31 days ago

When might makes right is your code of ethics, you lose your right to outrage.

rlittlesii.bsky.social•30 days ago

I'll cry for them, when they pay me for the code they've scanned of mine.

designpappa.bsky.social•31 days ago

Classic hypocrisy the U.S. government is known for. We can support the overthrowing of foreign governments, but when others do the same, it's terrorism. Not surprised one bit. Tech feudalism on full display.

matthewrdangle.com•30 days ago

I can not verify firsthand, but on Reddit, there are posts sharing screenshots where DeepSeek believes it's ChatGPT. I would not be surprised if they had trained their model from OpenAI.

That being said... I find it hypocritical that their data should be protected while ours is not.

chromaticnova.com•30 days ago

Hilariously ironic.

Kinda sucks if it’s true though, because it means the claims that Deepseek was cheaper to develop and train aren’t really legit, if they just piggybacked on top of existing genAI tech.

Comments

Posting Rules

Reply