David Friedberg alleges AI labs used his chat history to train later models
A post quotes Friedberg saying a later model, accessed through a different account, described the same novel scientific idea discussed in an earlier chat. He calls these “a handful of anecdotal experiences” but believes his team's work was used for training.
TLDR
David Friedberg, quoted in a user's post, alleges that ideas from his team's AI conversations resurfaced in a later model. He describes the experiences as anecdotal but concludes that the conversations or analyses were used for training. Friedberg calls the work his organization's intellectual property and says it had no NDA or confidentiality protections with the provider. He ties his interest in open source to that concern: he doesn't want chat logs used for training that could spread his organization's ideas to the wider market.
Combined views
401.4K
4 Sources, first seen ago
