Google Accuses Rivals of Copying Its AI Without Permission

Google is facing accusations of copyright infringement while simultaneously claiming protection over its own AI technology. The company has relied heavily on various online materials to train its Gemini AI models. This has led to several copyright lawsuits against Google and other firms in the AI sector for failing to compensate content creators.

Allegations of Intellectual Property Theft

Recently, Google accused what it refers to as “commercially motivated” actors of attempting to replicate its Gemini AI. The claims surfaced in a report where Google described being subject to “distillation attacks.” These attacks involve querying the Gemini model up to 100,000 times to extract its underlying principles, likened to copying homework in academia.

• Type of Attack: Distillation attacks aimed at intellectual property theft.
• Query Volume: Up to 100,000 times on the Gemini model.

This approach raises questions about the ethical implications of content use within the AI industry. Google labeled these attempts as a violation of its terms of service, highlighting a perceived double standard given its own practices of scraping content.

The Stakes in AI Development

As companies invest billions in developing AI capabilities, concerns about competitive advantage intensify. Differentiating products has become crucial as outputs from various AI models increasingly resemble one another.

In early 2025, the subject of model distillation triggered controversy when Chinese startup DeepSeek showcased a more cost-effective AI solution. This event prompted OpenAI, the creator of ChatGPT, to suggest that DeepSeek may have violated its terms of service, evoking criticism for hypocrisy from the online community.

Future Implications for AI Companies

Google’s Threat Intelligence Group has warned of more incidents involving intellectual property theft. Analysts believe that the methods for cloning AI capabilities could become more prevalent, as adversaries can leverage legitimate API access to replicate AI services.

• Key Concern: High exposure of AI models to theft.
• Attack Example: Attackers using over 100,000 prompts to mimic Gemini’s capabilities.

In response to these attacks, Google has reported proactive measures, stating that its systems recognized these intrusion attempts in real-time. This incident illustrates the vulnerabilities facing AI companies, especially as they aim to monetize their technologies through various revenue strategies, from subscriptions to advertising.

As the landscape evolves, smaller entities could replicate successful models, which raises additional challenges for established players like Google. The future of AI technology and its intellectual property rights continues to present a complex and contentious issue.