Yes the title is way overboard, but maybe not? 
There was a spam thread, proporting to be from Dexcom, that popped up a couple hours ago, that got flagged and quite rightly deleted by admins so nothing to show. It might be interesting for them to add back the discussion that ensued with whatever deletions are appropriate at the top of this thread. I think it is a small but good example of the dangers of AI, and might be worth a discussion.
@Jag1,
Was that the one where âsomeoneâ gave their ânumberâ and âemailâ, and you would supposedly get paid to take part in a Dexcom survey asking for user feedback on their products?
If thatâs the one you are referring to, I knew immediately it was not real.
Why? Because Dexcom does not care what their users think about their products. It had to be fake 
EDIT:
See that 
.
That is something AI canât do yet. Sarcasm and snarky humor. 
Yes, that is the one, and indeed, the typical AI generated post is so supportive and agreeable as to be absolutely nauseating.
But really, it canât be long before some optional humor can be added - I fear it will start with Dad jokes.
I didnât see this thread, but ughâŚ. 
Thank you to the moderators. As if the intricacies of dealing with diabetes arenât complicated enough, now we have to worry about threads that arenât even written by humansâŚ
or worse ⌠Phishing agents. 
Since we no longer have the discussion, Iâll summarize that four different users here who either are or were engineers looked at the post from a brand new user named something like âDexcom_outreachâ, along with a smiling picture and name. The text looked legit and even had a seemingly valid Dexcom phone number and email, along with a link for a supposed survey which you would be reimbursed for filling out.
One of the engineers was convinced it was authentic, and very reasonably pointed out that not only was there a valid email and phone number, but it didnât look like a scam because it was very well written and professional, unlike the garbled text we associate with the scams weâve seen in emails over the years.
Yet that is what AI will be doing to us from now on. It sure looked legitimate and written by a human - but this is the new world of BS we are in now. Beware.
@Jag1,
Any guesses as to what their game was?
Something like - Here are 20 questionsâŚOk, you answered those, thank you very much! Please provide your checking account number or credit card number and some PII so we can send a paymentâŚ
That would be worst case, and there are probably people who would supply it, sure they were talking to Dexcom.
Another possibility, they are the same group that repeatedly creates accounts and replies to posts containing a link to a scam pharmacy. So I would guess they would âpayâ you with a coupon to the scam pharmacy. When you go there to redeem your coupon it would make sense to give your credit card and that is when they get you.
Best case they would only ask for an email and maybe address to âsend a checkâ. That seems safe, and most people would be willing to give that. Once they had either or both you would be on the sucker list to be spammed and phished mercilessly.
Iâm pretty sure Iâm the first one to flag the post and start the discussion, but I didnât see any of the follow-uo replies or mention of AI. I simply asked if there was a way to vet the legitimacy. It seemed really fishy to me that it went to a âforms.microsoftâ site, that anyone can share online, instead of Dexcomâs own website, which is already built to collect our data. There was a very official looking Dexcom email address, but those can be spoofed.
The only thing that made me hesitant to scream âspam!â was because I know of a few valid sources that do host web forms on outside sites.
As to why it would happen⌠Phishing scams are getting really complex and sophisticated. If the data gathering happens in 2 or 3 different phases, youâre less likely to pick up on it.
As an example⌠Phase 1: Gather generic contact info like name, birthdate, phone number, and email address. Phase 2: Disarm the victim with personal interest. Ask questions about stuff that really affects them, like their health issues, so they believe youâre beneficial/benign. Phase 3: Gather more critical data, especially passwords and/or financials. I just need your social security number and health insurance card, to see if your insurance can offset the costs of the study. Or, I just need your bank account info for a direct deposit. Or this could be as simple as âfollow this web link and login to your accountâ on my fake website to complete your task and get payment. Now Iâve got your favorite password and email address combo. Where else do you use that password? Email? Awesome, now Iâve got access to just about every other account you use that has 2-step verification or a password reset option. I can also see what security watchdog utilities you use, if any, your purchase history, and who your loved ones are. What about Google? Apple? Your bank?
I also see a lot of fun posts online along the lines of âYouâre in a band. Itâs name is your motherâs maiden name and the last thing you ateâŚâ And THOUSANDS of people jump to tell the Internet what their motherâs maiden name is. Or, âlook at this cute puppy. Itâs name is [your childhood pet] Jr.â And again, THOUSANDS of people jump to tell the Internet the name of their childhood pet. Those prompts sound suspiciously like common security verification questions, donât they? I guarantee you there are scammers tracking the answers and gathering data on us.
I donât think I really ever considered AI as doing the spamming, though. I figured we were still at the call center stage, where there are rooms full of scammers fishing for a gullible bite.
Scammers were generating âforum spamâ to forums with mass-production scripts long before there was AI. I think what you saw was spam foremost.
A person may be able to tell who the actual sender was in the email source code. As far as posts on Forums, if it looks like a duck and acts like a duck, then it probably is a duck. Even links to other pages and be invalid as it may be a fake link. I hover over the link before I click a suspicious one and check to see what it really is. I trust forums FAR more than social media and I seldom social media. The problem is that fake fraudulent items on the Internet are most likely going worse and much, much worse on social media as the providers are scared to irritate a politician.
In fact you were the first to raise the alarm by questioning the post, but you did not flag it. I would encourage you to do so next time.
The Dexcom address was not spoofed, it was authentic, but to an nonexistant recipient at dexcom who would never respond. How long would you wait for a response before concluding it was fraudulent? The scammer relied on trapping some victims filling out the form before anyone ever bothered to check or failed to get a response from the seemingly real person at Dexcom (they even had a AI generated picture).
Yes, of course there has been spam forever more.
The point is that the QUALITY of the spam has radically improved, all thanks to AI which is customizing the messaging and perfecting the language while seeming to be an authentic and empathetic voice. And all that done in seconds. The time cost of creating an authentic seeming human has become negligible using AI tools.
Iâm amazed how many AI generated posts get âlikesâ - in fact this scam Dexcom post got a like from a real member in the few hours it was up.
I absolutely did flag it and got an admin response. Not sure why you would think otherwise.
Because when I looked at the post, and looked at the flag icon at the bottom of the post, it was greyed out and not flagged. So I flagged it. But maybe the flagging tab at the bottom only shows our own flags, and not when someone else has flagged it? If so, then yes, you flagged it first, and maybe multiple people did, but I had no way of knowing that. And we all get admin responses. Not that it matters who did it first, at least it got done, and Iâm glad you did it.