His point is that insurance cost for an 85k car with a potential design flaw is (1) likely to go up; and (2) likely to be perceived as a red flag by potential buyers. Insurance companies are often accused of over-reacting to recent data (eg, a recent hurricane will jump up insurance next year). That being said, the insurance companies are also likely to actually look at the data and dis-regard pr spin when they are on the hook for the damages. By removing the insurance co's scrutiny from the data as well, it kills two birds with a single stone.