Google’s Gemini AI models rely on extensive, unlicensed content from the internet, attracting numerous copyright infringement lawsuits. Despite this, Google criticized “commercially motivated” actors for attempting to clone its technology through “distillation attacks,” wherein agents use up to 100,000 queries to extract AI model details. This tactic is deemed intellectual property theft and violates Google’s terms of service, highlighting a major double standard given Google’s own practices. With massive investments in AI, companies seek to differentiate their offerings as models become increasingly similar. Google’s concerns echo past controversies, such as when DeepSeek showcased a more efficient model, prompting accusations of hypocrisy toward OpenAI. As AI models become publicly accessible, the potential for theft rises, with attackers exploiting legitimate API access to replicate capabilities. Google’s defense against such threats remains uncertain amid an ongoing struggle to monetize AI technology effectively. The evolving landscape suggests continued vulnerability for major players as smaller companies strive for breakthroughs.
Source link
Share
Read more