DeepMind researcher: AI firms turned humanity's free commons into trillion-dollar products

AlexTensor · x · 2026-11-21

Google DeepMind's ibab argues an under-discussed LLM ethics angle: pretraining distills humanity's collective work — writing and vast open-source code — into model weights, content released as a common good before anyone knew LLMs would exist. AI firms then trained on trillions of tokens and sell the result back for hundreds of billions. Reposting it, ChombaBupe adds that companies 'took something we released for free for the common good' and built trillion-dollar businesses, while claiming ownership of model outputs with no attribution.

Related event: Google Researcher Argues LLMs Distill the Human Commons, Weights Should Be Open-Sourced(4 posts)→

Original post →

More from AGI Musings

AGI Musings channel →