Google is reportedly relying on Claude, an AI model developed by Anthropic, to improve the performance of its own AI system, Gemini. According to a report from TechCrunch, contractors hired by Google will be hired to compare the responses generated by Gemini and Claude against the same user prompts. Your job is to evaluate the quality of the output based on factors such as truthfulness, clarity and detail.

The process works like this: Contractors are shown responses from Gemini and Claude for a specific prompt. They then have up to 30 minutes to carefully review each answer and rate how well it meets the assessment criteria. This feedback helps Google identify areas where Gemini may need improvement. Interestingly, the contractors noticed some unusual patterns during this evaluation process. Occasionally, Gemini’s works included allusions such as “I am Claude, created by Anthropic,” sparking curiosity about how closely related the two systems might be.

One thing that stood out in comparison was Claude’s strict approach to security. Contractors reportedly noted that Claude avoids responding to requests it deems unsafe while maintaining strict adherence to high ethical standards. On the other hand, Gemini sometimes flagged these unsafe prompts, but did so in a way that contractors found more detailed but perhaps less stringent. For example, in a case involving requests involving nudity and bondage, Claude refused to comply, while Gemini viewed the request as a serious security breach.

Google has a system for these comparisons and uses an internal platform that makes it easier for contractors to test and review multiple AI models side by side. However, the use of Claude for this purpose has raised some questions. Anthropic’s terms of service specifically state that users are not permitted to access Claude to build or train competing AI products without permission. This rule applies to other companies using the model, but it is unclear whether it extends to investors such as Google, which provides financial backing to Anthropic.

In response to the speculation, Shira McNamara, a spokeswoman for Google DeepMind, clarified the situation. She emphasized that comparing AI models is common practice in the industry and is essential to improving performance. McNamara firmly denied any claims that Google used Anthropic’s Claude to train Gemini, calling such suggestions inaccurate.

For now, the collaboration highlights just how competitive the AI ​​industry has become, with major players like Google using a mix of internal development and external partnerships to improve their models. At the same time, it also raises questions about ethical boundaries and how companies balance collaboration and competition.