Technology
Unsealed court filings detail OpenAI concerns over copyrighted training data
Unsealed court documents show OpenAI executives worried about public fallout over using LibGen datasets, according to Hacker News.
The short version
- Unsealed briefs from the Authors Guild lawsuit reveal OpenAI executives discussed public relations concerns over using copyrighted data from LibGen.[Hacker News]
- Internal communications indicate OpenAI employees expressed worry about potential public scrutiny on forums like Hacker News.[Hacker News]
- Filings allege OpenAI initiated 'Project Clear' in summer 2022 to delete LibGen files from its storage systems.[Hacker News]
Key facts
- Then-OpenAI Research Director Dario Amodei described LibGen as 'a bit sketchier' as a training set.[Hacker News]
- OpenAI researcher Sam McCandlish stated concerns about optics regarding public posts about using copyrighted data from a Russian website.[Hacker News]
- OpenAI executed an effort named 'Project Clear' to delete LibGen files from its systems in mid-2022.[Hacker News]
What remains uncertain
- The full legal impact of the unsealed court filings on the outcome of the Authors Guild v. OpenAI lawsuit remains undetermined.[Hacker News]
Sources
Outlet counts describe coverage, not independent confirmation. Reports may share a wire service or original source.