xAI reportedly trained coding models on Claude outputs for months before losing access
Reports in the AI industry suggest that Elon Musk's xAI may have used Anthropic Claude outputs for months to train its own coding models.
Reports in the AI industry suggest that Elon Musk's xAI may have used Anthropic Claude outputs for months to train its own coding models before access was cut off.
What reportedly happened?
According to the reports, the xAI team systematically used Claude outputs to improve its own AI tools focused on coding. The process allegedly involved private accounts and the Blackbox AI service to access competitor technology. These activities were said to stop only after Anthropic identified the pattern and blocked access, citing internal rules around API and data usage.
Wider context: internal change and compute economics
The situation appears to overlap with internal changes at xAI. Reports mention a smaller team responsible for model training and the departure of several key leaders. There are also surprising claims about compute allocation. A massive amount of computing power reportedly bought by Elon Musk for xAI's own model development is allegedly being rented out to competitors including Anthropic and Google. That could point either to a strategic shift or to operational friction inside xAI.
Why this matters for the AI industry
The story highlights several important issues in the fast-moving AI market.
Ethics and terms of service: using one model's outputs to train a competing model raises serious questions about ethics, intellectual property, and compliance with service terms.
Intense competition: the case shows how aggressive the race has become among major AI players, where access to high-quality training data is a critical advantage.
Strategy and resources: renting out compute rather than using it internally may indicate strategic repositioning or development problems inside Musk's company.
Many details remain unconfirmed and based on media reporting, but the situation still underlines growing tension around transparency and resource control in the AI industry. More can be found in The Decoder report.
Sources: - anthropic.com - arxiv.org - reddit.com - google.com
Source
the-decoderSeeing a similar issue in your company?
If this entry touches a process, dataset, or implementation problem you already see in your business, it is usually better to start with a short diagnosis than chase the next fashionable AI feature.
Related newsroom entries
The first good automation candidate is not always AI
In many small businesses, a straightforward process automation delivers more value than adding AI too early.
Why a newsroom should explain what a tech update means for a small business
The mere arrival of a new AI tool on the market usually does very little for the owner of a small business.