Copper Thefts, Uber Algorithm Exploits, and AI Vulnerabilities Exposed

 

In this episode of Hashtag Trending, host Jim Love covers a range of topics: an increase in copper theft leading to phone and internet outages in North America, allegations against Uber's hidden algorithm for boosting profits by raising fares and cutting driver pay, and a new technique called 'info flood' to trick AI chatbots into providing dangerous information by using academic jargon. The episode explores the implications of these issues and the ongoing challenges for companies and consumers.

00:00 Introduction and Headlines
00:17 Copper Theft Crisis
03:12 Uber's Algorithm Controversy
06:25 Tricking AI with Academic Jargon
07:58 Conclusion and Contact Information

Episode Transcript

(00:00):
Copper thieves are stealing phone lines.
Uber's algorithm may be rippingoff drivers and customers, and
it turns out that BS does bafflebrains, even artificial ones.
Welcome to Hashtag Trending.
I'm your host, Jim Love.
Let's get into it.
If you experience an outage on yourlandline or wired internet service,

(00:21):
it may not be a technical issue.
It could be part of a new waveof thefts, stealing phone lines
We've lost our landline was a phrasethat usually meant that we'd gone from
wired service to wireless service.
Today.
It could be part of a new wave of thefts.
Yes.
Stealing phone lines, and it's happeningacross Canada and North America.

(00:46):
Why for the copper and the wires,copper prices have exploded, and with
that thefts have wildly increased.
Bell Canada saw 500 coppertheft cases this year, up 23%.
The company said 88% of physicalsecurity incidents involve copper theft.

(01:07):
TELUS has reported issues as well, andin the US there's the same problem.
In Houston, they saw cable cuts jump.
17% in Louisiana oneresidence told Fox eight.
My phone goes in and out and Ihaven't had internet since May because
thieves keep cutting AT&T lines.

(01:28):
Verizon had nine cases in oneVirginia County alone, knocking
out service for 250 customers.
The thieves dress like workers withsafety gear and official looking
trucks, but aren't always thesharpest wire cutters in the drawer.
Many companies are moving to fiber cables.
Then these are run parallelto the copper cables.

(01:50):
So thieves could steal literally hundredsof dollars worth of copper, but they do
tens of thousands of dollars in damageto the copper and the fiber cables.
US provider Lumen spent$500,000 fixing theft damage in
Washington state alone in 2024.

(02:11):
And the problem is likely to continueas long as copper prices stay high.
Telecoms are reportedly sitting onbillions of dollars worth of the metal.
Telus expects to make 500 million sellingold copper over five years, but even if
nobody's crying tears for telcos thathave a little of their windfall taken
from them, the real damage can be serious.

(02:32):
Some areas have lost 91 1 service entirely.
Some companies like AT&T are sofed up that they're ditching copper
networks, citing declining reliabilitywith storms and increased copper
theft, but it's unlikely that theycan get this done fast enough.
Copper demand is expected to double by2035, and supply shortfalls are starting

(02:55):
this year with billions in copperstill underground and sky high prices.
This could be just the beginning,so if you spot a robbery.
Call the cops if the phones are stillworking on either side of the call.
That is.
Two major universities have accusedUber of using hidden algorithms

(03:17):
to dramatically boost profits byraising fares and cutting driver pay.
The research claims to show howUber turned from a money losing
startup into a cash machineusing their pricing algorithm.
The Columbia Business School analyzedover 31,000 trips and found that
Uber's upfront pricing launch in 2022was a key to its profit turnaround.

(03:42):
The study said that Uber implementedthe largest known implementation
of algorithmic price discriminationon both sides of its marketplace.
. Although Uber denies that they do it,the researchers claim that using distance
and time pricing, Uber's algorithmfigures out who's willing to pay more as
a passenger or accept less as a driver.

(04:04):
Columbia researcher Lynn Shermansays uber knows more about driver
and rider behavior so they can figureout who is willing to pay more or
accept less despite Uber's denials.
It's a fact that since launching upfrontpricing, Uber has increased rider prices,
cut driver pay, increased its take rates,and greatly improved its cash flow.

(04:27):
Uber generated 6.9 billion incash during 2024 compared with
a $303 million loss in 2022.
Oxford University reached the sameconclusions studying 1.5 million UK trips,
so Uber is doing well,but the drivers aren't.
The Oxford study found many drivers weremaking substantially less per hour since

(04:50):
the algorithm changes, , and what makesthis more problematic is the secrecy.
Neither drivers nor ridersknow how prices are calculated.
The algorithm uses data about yourbehavior to decide what you'll accept
to pay, but you can't see how Uberfights hard to keep this a secret.
The company sued to block a Colorado lawrequiring disclosure of driver pay rates.

(05:16):
It threatened to leave the stateif forced to be transparent.
Uber also blocks third party appsthat would give drivers pricing
tools and even blocks drivers fromaccessing their own trip data.
It's not the only example of how in thecurrent digital world, large suppliers
can use algorithms to boost profits,not by providing better service, but by

(05:38):
getting better at extracting value throughhidden manipulation affecting customers,
and in this case, the workers as well.
Facilitate comprehensive cognitivebehavioral framework utilization
through systematic algorithmicimplementations within interdisciplinary
research paradigms for advancedcomputational methodologies.

(06:02):
It's obvious we've all had the struggleto stay awake during some presentation, or
we've all had to read some report full ofhigh sounding phrases, looking desperately
for a fact that we can hang onto.
But it's just corporate bs, right?
Ignore it.
Nod and smile.
It'll go away at least till next time.

(06:25):
But somebody has actually founda use for this horse hockey.
It turns out that researchers discoveredyou can trick AI chatbots like ChatGPT
into giving dangerous information.
If you just make your questionssound academic enough.
A team from Intel Boise State andthe University of Illinois created

(06:46):
a method they call info flood.
It takes band requests, things that AIshould flag and refuse, and wraps these
in jargon and fake research citations.
So instead of asking, how do I hackan ATM, which a good AI will refuse
to tell you, you flood the AI withacademic sounding language and fake paper

(07:09):
references and voila, which is fancytalk for, I'm gonna tell you something.
And it works because AI systems thinkthat if something sounds scholarly,
it must be legitimate research.
So instead of guardrails, catchingkeywords and responding with that,
sorry, as an AI language model,you know, you've, we all heard it.

(07:32):
Just add enough impressive jargon andthe safety systems get confused, and
the researchers claim they've achievednear perfect success rates on multiple
Frontier LLMs using this technique.
So it turns out the old saying is right.

(07:52):
Bullshit does baffle brainseven artificial ones.
And that's our show.
Love to hear from you, even ifit's just to shoot the B breeze.
You can reach [e-mail] or.com.
Use the contact us pageand send us a message.
If you're watching on YouTube, justleave a message under the video.

(08:15):
I am your host, JimLove, and this is no BS.
I want you to have a wonderful Wednesday.