These are good points but: > JavaScript is not a great language to build high speed low resource inference engines. This is a browser plugin analyzing html do

These are good points but: > JavaScript is not a great language to build high speed low resource inference engines. This is a browser plugin analyzing html downloaded at human browsing speed. I doubt performance is a primary requirement. > Even if achieved in JavaScript, the model would be simple to pull out and reuse elsewhere, possibly to learn how to game it. The binary model is obfuscated but still distributed. I expect that the added difficulty of working with the binary model is small compared to the overall challenge of gaming it. > Good models will be built by good data scientists, whose favorite tools likely don't produce models that can be serialized for use in a JavaScript inference engine. Models built in research-optimized environments can be translated after the fact to match production needs. (It is getting easier with e.g. standard interchange formats for neural models.). Kapersky is a resource-rich org working on security software -- exactly the folks whom, if diligent and well intentioned, should invest in such hardening.