Matthew Sun argues that AI companies should stop calling themselves "labs". He is a software engineer and writer based in San Francisco. He says the word borrows the credibility of science for large commercial firms. Sun examined 100 recent research items on the OpenAI and Anthropic research pages. Most had no matching peer-reviewed paper at a journal or conference. He notes it is unclear whether any were submitted, rejected or still under review. Neither company answered requests for comment. He credits both for studying unflattering topics such as sycophancy and disempowerment. He also faults Anthropic for calling student cheating levels hard to measure. He wants firms using the label to open-source code, preregister studies and seek peer review.

• 100 research pieces examined
• 75%+ trust scientists (Pew 2026)
• AI spending near 2% of US GDP