text
stringlengths
3
159
Gosh,
I wonder why they aren't selling.
And researcher, and I guess fashion designer, Teresa Barton,
is recycling old Nvidia GT 730 graphics cards
into purses that cost $1,024 exactly,
which is one way to make enough money to finally afford an RTX 4080.
Barton also offers higher-end purses
fashioned out of H100 AI chips
for over $65,000.
These repurposed GPUs aren't functional, according to Barton,
they're ethically sourced from data-centred dumpsters,
but the fan still works,
which should mean your wallet and keys get excellent airflow.
And we'd appreciate more airflow this Wednesday when
the next episode of TechLink drops,
so come on back,
if you're a fan.Hahaha,
I'm sorry,
oh God, I'm so sorry, please
That was,
I didn't mean to make that joke,
that was not,
please.Please come back.
Welcome to TechLinked,
where we're going to patch your knowledge of the world
with some hot, fresh tech news.
Microsoft has sheepishly admitted
that the Windows 10 security updates it released between March and August
were in fact a serious downgrade that undid several old security patches,
like a bad case of scurvy reopening decades-old wounds.
Luckily, I've never been on a pirate ship.
And also, luckily, it seems that only a small number of systems were directly affected,
enterprise users still using the original 1507 version of Windows 10 released in July, 2015.
According to Senior Director of Threat Research at Immersion Labs
and also auto-generated character in Skyrim, Kev Breen,
Microsoft's statement points to an integer overflow vulnerability
where the build version numbers verified by the Windows Update service
fell into a range that caused a code error.
This then led to certain optional components
to be reverted back to their original unpatched versions,
i.e. exactly how they were when they were first manufactured.
Security experts have given this particular bug a score of 9.8 out of 10,
which would be great if this was an Olympic dive or a dressage performance. Mh-huh.
Microsoft, though, has now released a new patch
to patch the old patch that unpatched this particular patch of PCs.
The US Department of Justice's antitrust trial against Google started this week.
Oh, not that one.
This is a new one.
As you may know,
Google was recently found liable for illegally abusing the dominant market power of its search division
and is now awaiting the court's suggested remedies
to prevent it from continuing to do so in the future.
Ah, the courts.
Now, however,
the company is again in court facing similar accusations regarding its ad division.
As with most antitrust disputes,
a major point of contention in the case
is how to define the market that Google is supposedly dominating.
The DOJ says that Google is acting as an illegal monopoly by intertwining its various ad tools,
specifically its publishing software ad manager
and its ad exchange network, AdX,
and its advertiser network market, AdSense,
leading to website publishers to feel trapped.
Google, however,
says that it's part of a much bigger ad market
where it competes with fellow tech giants
like Amazon and Meta and Microsoft and Roku and TikTok and Yahoo,
the company.
That one, not just the feeling.
Of course, most of those supposed rivals
really only publish ads within their own walled garden,
whereas Google,
much like Jesus and the smell of microwaved salmon, is everywhere.
In other Googlish news,
Nevada's Department of Employment Training and Rehabilitation
has apparently paid Google over $1.3 million
for access to AI-powered cloud software
that will recommend whether or not
unemployed citizens should receive government benefits.
Hmm.
In the past,
organizations have typically relied on AI and other less sophisticated algorithms,
primarily to handle relatively low-stakes decisions that don't require much human oversight.
However, this system is apparently going to be used
to analyze evidentiary documents and transcripts for unemployment appeals hearings,
which are likely to be relatively complex cases.
I mean, just look at those words.
They're big words.
According to Nevada officials,
this system will help them eliminate the backlog of cases
that has existed since the pandemic
by reducing the time it takes to write a determination,
from several hours to just five minutes in some cases.
While the state promises each case will receive human review,
five minutes of human review isn't particularly reassuring.
This is yet another example of how automation is being used
as a way to place more and more responsibility
into the hands of unaccountable machines.
Similarly, an experiment by More Perfect Union
found that several Uber and Lyft drivers