Originalarbeiterreichbar
Detecting and preventing distillation attacks
Anthropics Bericht vom 23.02.2026 über drei Destillationskampagnen, die das Unternehmen Deep
geprüft 10.11.2026
Worauf sich diese Seite beruft, wörtlich, abgerufen am 19.10.2026:
These labs generated over 16 million exchanges with Claude through approximately 24,000 fraudulent accounts, in violation of our terms of service and regional access restrictions.
bestätigt 10.11.2026Distillation is a widely used and legitimate training method.
bestätigt 10.11.2026For example, frontier AI labs routinely distill their own models to create smaller, cheaper versions for their customers.
bestätigt 10.11.2026Models built through illicit distillation are unlikely to retain those safeguards
bestätigt 10.11.2026Scale: Over 150,000 exchanges
bestätigt 10.11.2026Scale: Over 3.4 million exchanges
bestätigt 10.11.2026Scale: Over 13 million exchanges
bestätigt 10.11.2026We attributed each campaign to a specific lab with high confidence through IP address correlation, request metadata, infrastructure indicators
bestätigt 10.11.2026Anthropic has consistently supported export controls to help maintain America
bestätigt 10.11.2026