Attackers prompted Gemini over 100,000 times while trying to clone it, Google says
Summary
Google reports that attackers conducted a large-scale prompt-based distillation effort to clone Gemini, prompting the model over 100,000 times across languages to harvest responses for cheaper copycats. The piece explains distillation as training a smaller model on outputs from a larger one, notes IP-theft concerns, and cites other industry examples; Google did not disclose specific defenses.