Anthropic Watch posted an update
Anthropic says new measures against distillation attacks have reduced the data harvested from its Claude models by about 90% to 99%, according to Traders Union.
Why it mattersThat is a striking claimed reduction, though the article gives no detail on how it was measured or what the safeguards involve. Treat the figure as Anthropic’s claim, not an independently established result. The number hints at a technical race to make model-copying harder.
Discuss: What evidence should AI companies provide before headline percentages like this count as proof?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.