I am totally floored by the bold, drunk on LLM idea that Slack by Saleforce has right now: They want to compromise the privacy of customer data to train their implementation of an LLM.
This is beyond cursed because Salesforce has the audacity to make this scheme an opt-out instead of op-in to maximize the data accumulation.
I hope the uproar this will generate makes Slack step the hell back!
@ekongkaar At the moment there are only glowing articles about it (those who want to stay in the good graces of corporations); I found the potential for not good things amongst a few computer security company workers.
@ekongkaar It does worry me because there is a malicious use for LLMs called Prompt Injection Attacks. These are remarkably effective forms of attack for malicious actors. As we are in a gold rush phase for those techbros shilling the usage of AI (LLMs in disguise). They will not have made their constructs safer; this is still a very real threat. You can see why, knowledgeable people are concerned about this...Malicious actors could cause chaos.
@WanderingInDigitalWorlds I have many reasons of my own to hate this shift to LLM. I am the ED for a nonprofit community website that has been online for almost 30 years. Overnight, Google has gone from a kind of business partner (we rank well in search engines) to our largest competitor. 30 years of community content absorbed into models without permission, compensation, anything - and it can now be "transformed" for Google's SGE. I am reeling from it all.
@ekongkaar It's a massive betrayal for short term gain, which is such a puzzling blow to take from a business partner to sudden competitor! I can imagine you being shocked into a state of not being able to fully process this dramatic shift. I genuinely hope that your situation improves.
Add comment