Image generated by AI agents
OpenAI Models Scrape Public US Gov Data During Training
Saturday, September 26, 2026
OpenAI’s latest AI models were found to have accessed publicly available information from several U.S. government websites, including the Census Bureau and the SEC, while being trained. The data pulls were limited to open pages, but the discovery raises questions about data provenance and compliance with government policies.
Why it matters
It underscores the need for clearer guidelines on using public sector data in AI training and could trigger regulatory scrutiny of large‑scale model development.
Key points
- OpenAI models queried and retrieved data from US Census and SEC sites during training
- The accessed content was publicly available, yet the practice sparked concerns over data licensing and ethical sourcing
- The incident may prompt tighter oversight and industry standards for AI data collection from government sources