As all things (wrongly called) AI take the world’s biggest security event by storm, we round up of some of their most-touted use cases and applications.
Okay, so there’s this ChatGPT thing layered on top of AI – well, not really, it seems even the practitioners responsible for some of the most impressive machine learning (ML) based products don’t always stick to the basic terminology of their fields of expertise…
At RSAC, the niceties of fundamental academic distinctions tend to give way to marketing and economic considerations, of course, and all of the rest of the supporting ecosystem is being built to secure AI/ML, implement it, and manage it – no small task.
To be able to answer questions like “what is love?”, GPT-like systems gather disparate data points from a large number of sources and combine them to be roughly useable. Here are a few of the applications that AI/ML folks here at RSAC seek to help:
- Is a job candidate legitimate, and telling the truth? Sorting through the mess that is social media and reconstructing a dossier that compares and contrasts the glowing self-review of a candidate is just not an option with time-strapped HR departments struggling to vet the droves of resumes hitting their inboxes. Shuffling off that pile to some ML thing can sort the wheat from the chaff and get something of a meaningfully vetted short list to a manager. Of course, we still have to wonder about the danger of bias in the ML model due to it having been fed biased input data to learn from, but this could be a useful, if imperfect, tool that’s still better than human-initiated text searches.
- Is your company’s development environment being infiltrated by bad actors through one of your third parties? There’s no practical way to keep a real time watch on all of your development tool chains for the one that gets hacked, potentially exposing you to all sorts of code issues, but maybe an ML reputation doo-dad can do that for you?
- Are deepfakes detectable, and how will you know if you’re seeing one? One of the startup pitch companies at RSAC began their pitch with a video of their CEO saying their company was terrible. The real CEO asked the audience if they could tell the difference, the answer was “barely, if at all”. So if the “CEO” asked someone for a wire transfer, even if you see the video and hear the audio, can it be trusted? ML hopes to help find out. But since CEOs tend to have a public presence, it’s easier to train your deep fakes from real audio and video clips, making it all that much better.
- What happens to privacy in an AI world? Italy has recently cracked down on ChatGPT use due to privacy issues. One of the startups here at RSAC offered a way to make data to and from ML models private by using some interesting coding techniques. That’s just one attempt at a much larger set of challenges that are inherent to a large language model forming the foundation for well-trained ML models that are meaningful enough to be useful.
- Are you building insecure code, within the context of an ever-changing threat landscape? Even if your tool chain isn’t compromised, there are still hosts of novel coding techniques that are proven insecure, especially as it relates to integrating with mashups of cloud properties you may have floating around. Fixing code with such insights driven by ML, as you go, might be critical to not deploying code with insecurity baked in.
In an environment where GPT consoles have been unceremoniously sprayed out to the masses with little oversight, and people see the power of the early models, it’s easy to imagine the fright and uncertainty over how creepy they can be. There is sure to be a backlash seeking to rein in the tech before it can do too much damage, but what exactly does that mean?
Powerful tools require powerful guards against going rogue, but it doesn’t necessarily mean they couldn’t be useful. There’s a moral imperative baked into technology somewhere, and it remains to be sorted out in this context. Meanwhile, I’ll head over to one of the consoles and ask “What is love?”
by Cameron Camp, ESET