Speaking about AI…
I just read a really interesting article in Apple News, from The Atlantic, on ChatGPT Images 2.0 release. The author made a number of fakes, including driver’s license, a CVS receipt, an Uber receipt, etc. She showed her findings to OpenAI asking them why their “guardrails” weren’t working. She didn’t get a very good response. She also showed Google and they told her about a tool called SynthID, which worked pretty well at identifying fakes made with Google AI.
If you are able I would encourage you to read the article, even just by stopping by the library you may be able to access it. I don’t know if it is available in the print edition.
My thought on posting this was more awareness for us, but also it occurs to me we may have a new career opportunity here. I believe businesses, banks, etc., large and small will need to implement some kind of program to find, confirm, and react to the “bad actors” out there. The ones who will target businesses, scam old people, and on and on. It’s not hard to identify slop when an image has six-fingered people, but the fakes are getting so real, and OpenAI says it is easy to delete or manipulate the meta data.
Thoughts?