Rummaging through my work email tonight (had to submit timesheet and look for travel charge codes) and I found a gem from our IT department regarding AI "features" that our vendors are inflicting on our organization effectively without consent. IT has managed to disable Adobe's AI tools from appearing by default at the corporate level but this still leaves users individually vulnerable to accessing it via direct "Try this!" marketing from the desktop application.
A large portion of our company's workers handle Export Controlled Information (ECI) which we must keep secure from those not authorized to receive it, lest we be put in Federal prison. Ignorance is no excuse.
Over and over we have seen that LLMs absorb and leak sensitive information. Securing LLMs is not trivial. Further, what happens when an LLM hallucinates a response that can be considered export-controlled? Those so dedicated to pushing these tools out without any thought will be the first to throw users under the bus.