27 August 2026

Anthropic shares Claude data and watermarking research with universities

First reported

Ahead of AI ran this on , 5 days before the next source picked it up.

  • Anthropic, the company behind the Claude chatbot, gave Stanford, Oxford, and METR access to conversations Claude has had with users for academic study.
  • Early research found that more than half of these conversations involved high-stakes decisions where users sought legal or financial advice from Claude.
  • Anthropic also revealed a new watermarking process that marks Claude's text outputs so researchers and users can identify when text came from the AI rather than a human.
  • One newsletter focused on how the watermarking actually works technically, while the other emphasized the research access and what conversations showed about how people use Claude.

Where they differ

  • The Rundown AI

    led with Anthropic's decision to open user conversation data to researchers.

  • Ahead of AI

    prioritized the watermarking announcement and its technical explanation, suggesting different views on which development matters more to readers.

What each one reported

The Rundown AIRowan Cheung

Anthropic granted Stanford, Oxford, and METR access to Claude chat data for research, with early findings showing over half of conversations involve high-stakes tasks like legal and financial advice.

Ahead of AISebastian Raschka

Anthropic announced a new watermarking process for Claude model text outputs to identify AI-generated text. Raschka provides an in-depth explanation of how the watermarking technique works under the hood, arguing that understanding the mechanism from scratch is useful for both users and researchers to evaluate the pros and cons of this approach.