#trainingdata

The problem: a training corpus can amplify false facts A language model does not know...

A September Authors Guild motion cites discovery evidence that OpenAI staff knew LibGen was risky and used LibGen-derived datasets, while leaving infringement, fair use, willfulness and Microsoft liability unresolved.

A plain-English guide to whether AI companies can legally train on your data — the US fair-use rulings, the UK Getty judgment, the EU opt-out and the US Copyright Office's warning

Originally published at norvik.tech Introduction Explore the implications of Snorkel...

AfterQuery went from a $300 million startup to a $3.2 billion company in five months — a tenfold jump that makes it the fastest launch-to-unicorn in Y Combinator history. It doesn't build a chatbot. It builds the training data every chatbot runs on.

In its 8 September write-up of the Navier-Stokes result, OpenAI stated that while no specific user data was accessed to solve the problem, it 'cannot rule out' that de-identified data from two mathematicians' use of its products helped improve its mo

AI models memorise fragments of their training data and can be prompted to reproduce them, turning membership inference and data extraction into a genuine privacy risk. Here is how

Twitch added a switch to stop Amazon training generative AI on your streams. Its own product chief explained on stream why it defaults to on, and that answer tells you what the setting is really for.

Stack Overflow received 1,490 new questions in July 2026, down from 6,414 in July 2025 and 176,610 in July 2014 -- a 118-fold collapse in the public programming corpus that trained today's coding models.

A hack of AI music generator Suno exposed source-code files and comments naming YouTube Music, Deezer, Genius, stock libraries and podcast RSS feeds as data collection targets, the most specific provenance evidence yet in the music industry's copyrig

Anthropic's physical book scanning program isn't primarily a story about destruction. It's about what happens to knowledge after it's captured — who controls it, who can verify it, and what happens when the organization holding it closes.

DataClaw0 turns the grind of cleaning and labeling training data into a learned skill -- a small model that refines raw, messy multimodal streams into dense, purpose-built lessons.