← @kwm Twitter archive

Keith

@kwm

🛑 A misuse of machine learning for threat detection: Because we can accurately model that which is likely to be good, the product suggests “that which isn’t known good, is probably bad.”

💡 That which isn’t known good is statistically far more likely to be *new* than bad.

9/10/2020, 12:29:38 PM

Favs: 34

Retweets: 12