Search
Anthropic States That New Protective Measures Have Substantially Curbed Data Theft Through 'Distillation Attacks'
6 day ago /
Read about 0 minute
Author：小编
On October 13 (local time), Anthropic made an announcement revealing that its newly implemented protective measures have notably diminished the likelihood of competitors pilfering data from the Claude AI model via 'distillation attacks.' These measures have led to a remarkable 90% to 99% decrease in the volume of stolen data. Moreover, the new safeguards have made detection more challenging, thwarted extraction endeavors, impeded large-scale distillation operations, all while keeping the false positive rate under 0.1%.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
