When safety features inadvertently encourage riskier behavior…
“AI safety features may actually decrease value by encouraging risk-taking that overwhelms the safety benefits.”
The Peltzman Effect posits that safer AI might lead users to engage in riskier behaviors, potentially offsetting safety gains due to a perceived reduction in danger. This behavioral adaptation creates new challenges.