Google has been accused of using Anthropic's Claude output to improve its Gemini model, raising questions about licensing agreements. The controversy surrounds a leaked communication from Google's outsourced team, which showed that the company was comparing the responses from Gemini and Claude to determine which was better. However, Anthropic has explicitly prohibited the use of its output for building or training competitive products, which has sparked controversy over whether Google has obtained the necessary authorization.
According to the leaked communication, the outsourced team was unaware of the source of the data at first, but later discovered that the responses were marked as coming from Claude. Industry insiders point out that while comparing outputs is a common practice in AI development, using a competitor's output without permission may raise ethical and legal issues. Google has denied using Claude for training, but has not responded to whether it obtained permission from Anthropic.
This incident highlights the importance of regulation and transparency in AI competition.
[Source: bluepoint.net](http://bluepoint.net)