Disrupting a coordinated model-distillation campaign

Worauf sich diese Seite beruft, wörtlich, abgerufen am 19.10.2026:

• This activity is consistent with adversarial distillation: the systematic and unauthorized use of one modelbestätigt 10.11.2026
• outputs or reasoning to help train, reproduce, or improve another model.bestätigt 10.11.2026
• high-volume spikes on July 24 and 25 consisting of 16,000 requestsbestätigt 10.11.2026
• using a relevant extraction pattern from over 4,000 usersbestätigt 10.11.2026
• Further investigation identified related prompt-pattern activity across a cluster of more than 15,000 users, which we fully disrupted by July 28.bestätigt 10.11.2026
• we attribute a core cluster of the activity to individuals associated with Moonshot AI, the developer of Kimi.bestätigt 10.11.2026
• These figures describe attemptedbestätigt 10.11.2026
• Extracted reasoning could be used to train another model without preserving the safeguards applied to the original modelbestätigt 10.11.2026