Michael Flux(@michaelflux):Anyone serious about their data will run their own (or open source) models on their own hardware. Otherwise, too much risk that your proprietary info makes it into "training data" for everyone else. Plus the risk that some providers' model doesn't want to talk to you about your own data because it crosses some arbitrary line.

2026.09.14 07:50

Anyone serious about their data will run their own (or open source) models on their own hardware. Otherwise, too much risk that your proprietary info makes it into "training data" for everyone else. Plus the risk that some providers' model doesn't want to talk to you about your own data because it crosses some arbitrary line.