as someone who spent several years attacking ML models for security research, so does this > Around 30% of the applications Harmonic Security has seen being us

as someone who spent several years attacking ML models for security research, so does this > Around 30% of the applications Harmonic Security has seen being used train using information entered by the user. > That means the user's information becomes part of the AI tool and could be output to other users in the future. > … Firms will be concerned about their data being stored in AI services they have no control over, no awareness of, and which may be vulnerable to data breaches.