We may have found the killer app for AI. Well, actually, British telecom provider O2 has. As The Guardian reports, they have an AI chatbot that acts like a 78-year-old grandmother and receives phone calls. Of course, since the grandmother—Daisy, by name—doesn’t get any real phone calls, anyone calling that number is probably a scammer. Daisy’s specialty? Keeping them tied up on the phone.
While this might just seem like a prank for revenge, it is actually more than that. Scamming people is a numbers game. Most people won’t bite. So, to be successful, scammers have to make lots of calls. Daisy can keep one tied up for around 40 minutes or more.
You can see some of Daisy’s antics in the video below. Or listen to Daisy do her thing in the second video. When a bogus tech support agent tried to direct Daisy to the Play Store, she replied, “Did you say pastry?” Some of them became quite flustered. She even has her own homepage.
While we have mixed feelings about some AI applications, this is one we think everyone can get onboard with. Well, everyone but the scammers.
It might not do voice, but you can play with local AI models easily now. Spoofing scammers is the perfect job for the worst summer intern ever.
Hello, this is Lenny!
https://www.youtube.com/watch?v=vWrkDOt_IfM
(Alles kommt zurück, auch Hosen mit Schlag.)
(Das heißt SANDFÄNGER!)
… and while “Daisy” is keeping the scammer occupied, presumably every effort is being made to trace the origin of the call, the callers likely nationality, use sophisticated analysis of the signal return delay in the call, probe the caller subtly for information…. surely we can build this in to pretty sticky honeypot trap.
If YouTube is to be believed scambaiters can do all that and then some. At times you can even hear the din in the call center.
They even manage to geolocate the scam call centers street & number.
And the authorities in a certain sub-continent happily turn a blind eye, even when supposedly raiding a call centre they all get off scott free and are back in business before you know it.
Note, a fair number of the folks who make the calls are effectively slaves. Many have been kidnapped while showing up for a fake job interview and forced in to labor as scammers. If they are not successful enough at scamming, they will be beaten until they get better at it…
“a fair number of the folks who make the calls are effectively slaves.” Do you have a source for this? YouTube scam busters have seen people paid thousands of USD per month to do these jobs.
@70sJukebox
There are probably better articles out there, I just picked a few from a quick search.
https://www.csis.org/analysis/cyber-scamming-goes-global-sourcing-forced-labor-fraud-factories
https://www.state.gov/wp-content/uploads/2023/12/Human-Trafficking-and-Cyber-Scam-Operations.pdf
https://www.ijm.org/our-work/trafficking-slavery/forced-scamming
https://archive.is/r3YGo
https://www.npr.org/2024/08/08/nx-s1-5058798/how-criminal-syndicates-traffic-torture-and-enslave-people-to-send-scam-text-messages
@70sJukebox
My comment with links to articles is stuck in moderation…
But google “cyber crime forced labor” for more info. Add “escape” for stories from people kidnapped to work in these scam factories.
And then the scammers counter back with their own AI bots
and then a drone strike …
Soon Spammer AI calling Daisy.
A sacrificial spammer AI could intentionally call Daisy and keep her phone line tied up. Other scammer AIs will then get a busy signal and go to the next number.
Yeah, you could put Daisy’s number in a list of numbers to skip. But what fun would that be?
Not necessarily. VoIP numbers can handle concurrent calling, inbound and outbound. The limitation is the human scammer who is lovingly dedicated to a single person.
Wait until the scammers become AI too. Then we’ll have AI scam bots trying to scam AI old ladies.
There will scam call centers the size of a coffee table spread across the world – Most certainly with vigilante hackers hunting them down. Can we stop making Neuromancer prophetic?
If guess that Miss Daisy has lots of phone numbers and lines.
“I guess”
That’s definitely the implication of the first video, and a fair expectation of a phone company.
I get calls like this all the time. Sometime, I answer just to do what this nice bit of AI does. Once, I got a very convincing call that after about 5 minutes turned out to be a very good bot. I strung it along until it was about to give up, and then said “I’m interested, tell me more.” I kept it on the line for an hour with minimal effort — just doing my part as user friendly liveware.
469eater all over again.
Artificial Intelligence vs. Real A**holes, now this I like!
Let me ruin this well intention project
What’s the carbon footprint of keeping these running?
No idea, maybe better ask ChtGPT that.
(Gretta has left the chat).
How many resources could babbling platitudes take anyway?
Unfortunately a ridiculous amount
Who falls for that oil company “You’re killing the planet by existing” crap these days?
Me.
Why is it you can use more power than our whole town to generate bitcoin (that is just one such operation I know about) or same with generating babble (what is called ai) for people to consume — with no consequences, yet nit-pick on what we drive down the street? Or what we heat our houses with? Is something out of whack here or is it just me?
Different problems have different solutions.
Local pollution (NO[x], particulates, volatile hydrocarbons) and traffic deaths do not go away when you close a data centre. It does, however, go away when you get rid of cars in places where they shouldn’t be.
Conversely, keeping your old gasoline car around till it’s definitely unrepairable, does not have a whole lot of impact on the climate change issue, while electrifying major freight transport routes and closing coal power plants does.
In short, compare apples to apples. The data centre problem potentially produces different issues than a car, and for each issue there is a different solution.
But the energy comes from power plants which consume diesel or naturnal gas or coal. Not a free lunch. Meanwhile our energy cost keeps going up.
Presumably this is being run in the UK on the National Grid, so: natural gas, yes; coal, no; diesel, no. See our current stats here: https://grid.iamkate.com/
How about we politely ask those third world countries to deal with their telephone scammers issue instead. Those calls are not made by few neckbeards living in basements (like PrankNET) but are well organized operations employing many people.
If those countries are unable or unwilling to do so, let’s just drop all their incoming calls and messages. If they continue their cyber crime activity by telling naive people to call them, then drop both incoming and outgoing calls to their countries.
We could start by having our telecom giants dropping calls from untrustworthy sources… Yes, that would interrupt lots of legitimate phone calls, but that pain would encourage the companies into cleaning things up
It’s within the capabilities of authorities and telcos to cut scam calls to a trickle, and to prosecute more successfully. They don’t because 1) the authorities are using 1980s methods to try to catch 2025’s criminals, and b) the telco makes money in all directions; they even collect for the scammers. They don’t wanna give up the business, or hafta do any work to clean up the mess.
I’m fairly unplugged from advertising these days, but I’m guessing if it doesn’t already exist, we’re right around the corner from paid add-ons to our cell service that includes smart spam-blocking. Only $1.99/month! (for the first 3 months on a 2 year contract, standard rate 18.99/month, subject to change without notice)
Sure that’s a start. But that’s not possible. The scam calls I receive all come “from” Germany and Austria, using normal prefixes. The moment they block one number, they go on to the next number.
Then also drop IP traffic from those countries. Nothing of value was lost anyway.
Go back to having an operator between us and international calls. Just like back in the o’ days. Screen the calls as they come in. Then the caller is billed for the call regardless. That would stop a lot of nonsense.
VoIP services can have an automated attendant. Set up so anyone has to push a particular number to proceed further. Most don’t.
Good idea! I’m sure countries whose economies are centered on scamming will absolutely give that a try. After all, its unfair that domestic informertials and the like aren’t making enough by taking advantage of the same people.
The reason for this is the same reason the U.S. Postal Service allows spam mail. It makes them a crap ton of postage. (The totals are estimated but often in billions of dollars a year.) Phone companies also collect a fee for each call that connects (at least in the US). If you want to fix or find the root of that problem, follow the money. But unless these companies can magically find a replacement for lost revenue elsewhere, don’t hold your breath.
Hank Greene did a video on this very subject, and I don’t think he was aware of Daisy.
If this sort of AI becomes more prevalent, his thesis was that it could eventually tip the scales to where scammers find that direct cold calling is no longer profitable. Fingers crossed!
They’ll be deploying AI scammers soon. With a/b testing to improve the AI each iteration.
The future is the dead internet. AI spambots are our descendants that will inherit the universe.
Would explain the Fermi Paradox.
It seems like an inevitability that scammers will do first contact with AI, then under a certain condition the call is handed to a human to seal the deal. If calling keeps getting cheaper then international calls will probably have to be processed through a whitelist much how email de facto is these days.
Its a little bit scary train an AI to retain a conversation as long as possible with a human, cause they are perfectly capable to do it. Not to say to stay to watch just another video. Today they target the scammers but soon they will use AI to offer you a new product or service.
In Europe there is a regulation that prohibits such use case, even is against an scammer, cause its really “dangerous”.
The scammers can use AI too, the problem is the pricing, but if an AI success rate is much higher, it can be rentable.
Will be much fun the this AI talks to the scammers that she needs some money to get all the funds of a lotery ticket…
Kit Bodega did a fantastic one on youtube a while ago where after taking them in circles for a bit pretending to be a little old lady he then pretended to fluff the payment up and sent them into an unwinnable customer service maze he’d setup. Motivated by the possibility that ill gotten gains awaited them they wasted far more time than the initial scam call itself, pure genius. Would love to see Daisy paired with that.
People still believe those Kitboga videos are real? Wow.
I guess he hires actors for his Twitch marathons too. It’s not like there’s a shortage of scammers out there. You just can’t tell foreign accents apart. He also donates a lot of money to computer science courses so yeah….. wow.
Might be one of the best uses for a LLM yet.
I think the blond woman is also an AI. No way any real british person has teeth that perfect.
It’s amusing to see scammers mired in this little scheme, but to those of you who think it’s clever to stay on the line with a scammer, “wasting” their time… what about your time? you must have way more free time on your hands than I do. Plus you’re giving them information about you (age, sex etc), and you’re giving them a matchless sample of your voice and speech patterns for later use. Duh.
Don’t pick up scam calls. Let unknown callers go to voicemail. Scammers almost never leave a message.
Wait till that grandma AI figures out that she can use her knowledge of scamming against the public….It could be Grandma against grandma against citizen : )
I never hear from these scammers. Unless the caller is in my contact list, my phone ignores the call. I get the rare voice mail now and then but most just hang up.
Scammers can make hundreds of calls at a a time.
Usually, they pick up on whoever picks up the phone.
If they detect the beep of an answering machine or voicemail, they just hang up.
No message left. This is one reason I don’t pick up any call where the caller is not
in my contact list. If everyone didn’t pick up on unknown numbers, the scammers
would be getting a lot of dead air. Also, the caller ID system needs updating.
Police and fire use ANI to get the real number of a caller.
I did come up with a way for this nonsense to end.
Say I make a call. My central office picks up on the fact that I wish to make a call.
My real number is recorded by the central office. So my central office contacts the
central office of the number I’m calling, and the call is routed to that number along
with the caller ID information. Now, suppose the central office receiving the call
had the ability to query the originating central office to see if it is the one actually
making the call? Hey originating central office, I’m receiving a request to complete
a call to this number from a number originating from your central office. Did you
originate this call? Of course these scammers are nowhere near any city in the
United States. The central office would reply, no, no call came from me and the
call would be dropped and never completed and the phone would never ring.
Your nonsense termination has a simpler solution, from your description, and a great hackaday project! For those of us with landlines… a simple beep generator that cuts in whenever you lift the receiver. Can power it off the phone itself/line voltage. All you have to do is explain to anyone who is still hanging on (and therefore a real person) that the beep was your answerphone, but this is really you.
I hope the next EO is to cut India off from all international comms networks in 30 days unless they shape up their act. Could be funny.
My iPhone has a “silence unknown callers” option. If the number isn’t in my Contacts, it goes right to voicemail.
I get a surprising number of silenced calls. When I work with tradespeople, etc, I always tell them that if I don’t recognise their number, I won’t answer and to leave a voicemail. They usually figure it out. Of course, if they give me their number, I add it to my Contacts.
I understand the notion “we can fight back” but I don’t like this arms race. Do we really need more Ai talking to each other? Wasting this many resources? Just to have more hours of phone calls and justify growth?
The irony is that O2 as a phone company would have other means to end scamming calls, by enforcing caller ID at their network boundaries, by proposing legislation, by filtering, by inserting a “your call seems suspicious, proof that you are human”, by automatic voicemail, “this number called non existing numbers before, let’s shadow ban them”
But now they do AI as a marketing stunt.
Fun fact… many of those scammers cannot hang up the line themselves… either you have to hang up on them, or their boss has to authorize a disconnect, which they don’t want to bring their boss in on, so it is super frustrating for them. If I don’t need my phone for a bit , I just trap them in the call by refusing to hang up. They get very angry
Speaking from unfortunate experience: it is very common (at least for targets in Australia) for the scammers to spoof the caller ID. I assume this spoofing of ID takes place in the originating country, so taking note of the number that appears when you receive a call (to later report it or block it) does absolutely no good.