Warning for Windows 11 Users, Lyft Privacy Scare, and AI's Memory & Dolphin Communication

In this episode, Jim Love discusses Microsoft's critical warning for Windows 11 users about the removal of the inetpub folder, a startling Lyft privacy incident where conversations were transcribed, and the implications of ChatGPT's new memory feature. Additionally, he explores the potential of AI to decode dolphin language through Google's Dolphin Gemma model.

00:00 Microsoft Warns Windows 11 Users
01:57 Lyft Privacy Concerns: Are Your Conversations Being Recorded?
03:45 ChatGPT's Memory: What Does It Know About You?
05:55 Decoding Dolphin Language with AI
08:18 Conclusion and Sign-Off

Episode Transcript

(00:00):
Microsoft warns Windows 11 users.
Deleting inetpub folderdisables critical security.
Is Lyft recording your conversations?
ChatGPT has a memory.
It remembers what you asked it evenif you didn't ask it to remember.
And finally, do dolphinshave a real language?
Maybe large language models cantell us This is hashtag trending.

(00:23):
I'm your host, Jim Love.
Let's get into it.
Microsoft has issued a warning toWindows 1124 H two users regarding
the deletion of a seemingly emptyin net pub folder created by the
April 25th cumulative update.
Despite appearing redundant, thisfolder is essential for a security
fix implemented in the update.

(00:45):
Users who have removed it areadvised to restore the folder
to maintain system security.
The in net pub folder typicallyassociated with Microsoft's Internet
Information services or IIS webservers, was unexpectedly created
on the system drive of many users,including those without IIS installed.

(01:07):
Its empty state led some to believe itwas unnecessary prompting its deletion.
However, Microsoft has clarifiedthis folder is integral to a security
patch applied during the April update.
Removing the Pub folder compromisesthe effectiveness of the recent
security update, potentiallyleaving systems vulnerable.
Microsoft emphasized the importance ofthis folder in maintaining the intended

(01:30):
security posture of Windows 1124 H two.
Now those who have deleted theinetpub Pub folder should take
immediate steps to restore it.
This can be achieved by reapplyingthe April, 2025 cumulative update,
or manually recreating the folder.
Best to check with Microsoft'sinstructions to find out exactly what

(01:50):
they do, but ensuring the presenceof this folder is crucial for the
continued protection of your system.
A recent incident in Toronto hasraised significant privacy concerns.
After a Lyft passenger discovered thather private conversation during a ride
was transcribed and sent to her phone.

(02:10):
The event which occurred in March involveda woman who received a text message
containing a transcript of her discussionwith a friend during a Lyft ride.
The message which was sent from anunfamiliar number included a note stating,
this is what you said in your Lyft ride.
The woman expressed her alarmover the incident stating I

(02:31):
was shocked and disturbed.
It felt like a violation of my privacy.
She reported the matter toLyft, but the company has yet
to provide a clear explanation.
Lyft representatives have acknowledgedthe incident and are investigating
the matter and emphasizing they donot have a feature that records or
transcribes passenger conversations.

(02:51):
Privacy experts have weighed inon the situation highlighting the
potential implications for userprivacy in ride sharing services.
David Fraser, a privacy lawyer,noted if this was an experiment
or test by the company.
It raises serious, ethicaland legal questions about
consent and data protection.
The office of the Privacy Commissionerof Canada has been notified and is

(03:13):
currently assessing the situation.
They have urged any individuals whohave experienced similar incidents
to come forward and report them.
The incident underscores the importanceof transparency and consent in the use of
technology that may impact user privacy.
It also points out that sometimes wehave conversations in places that we

(03:33):
think are private but really aren't.
As investigations continue, passengersare advised to remain vigilant and
report any suspicious activitiesto the appropriate authorities.
and speaking of listening, whatdoes your AI know about you?
If it could remember and it can OpenAIannounced that ChatGPT would be able

(03:56):
to remember old conversations thatyou didn't ask it to save open ai.
CEO, Sam Altman said on X that thechatbot can now reference all your
past conversations and that theupdate aligns with the company's
goal to develop AI systems thatget to know you over your life.
It was first rolled out to those withthe Prime $200 a month subscription.

(04:19):
It has now appeared to the $20 amonth subscription, at least in my
ChatGPT, and I wanted to get someidea of what it actually knew about
me from our past interactions.
So I asked it a couple of questions.
ChatGPT.
What do you know about me?
You are a writer and podcasteryou have a keen interest in AI,

(04:40):
cybersecurity and the broader techindustry, often exploring complex topics
with a focus on practical insights.
You care about constructivefeedback and continuous improvement.
You're also not afraid to be direct whensomething doesn't meet your standards.
Overall, you combine creativity,precision, and a commitment to
quality in everything you do.

(05:00):
If there's anything else you'dlike to know or if you want to dive
into a specific topic, just letme know what is negative about me.
From what I've observed, a fewpotential challenges might come up.
Your high standards and attention todetail, while great for quality, could
sometimes lead to frustration if thingsdon't meet your expectations right away.

(05:23):
This might make collaborationwith others a bit more intense,
especially if they don't share thesame urgency or level of precision.
Another point could be that yourdecisiveness and strong opinions
might come off as a bit rigid orleave less room for flexibility.
While this helps in maintaining highstandards, it might also make it harder
to compromise or adapt when needed.

(05:44):
Ultimately, these traits are just the flipside of your commitment to excellence.
If you ever want to chat more aboutthis or explore ways to balance
those traits, I'm here to help.
And finally, do dolphinshave a real language?
Most studies to date would tell you no.
While dolphins can learn symbolsand appear to exchange different

(06:06):
sounds, linguists would saythey lack a large vocabulary.
Don't display a grammar or syntax or anyreal abstract thought, but AI may give
us a final and definitive answer on that.
Google has introduced Dolphin Gemma,an AI model developed in collaboration
with Georgia Tech and the Wild DolphinProject, aiming to decode dolphin

(06:29):
vocalizations and facilitate two-waycommunication between humans and dolphins.
Dolphin.
Gemma is trained on decades of underwateraudio and video data collected by
the Wild Dolphin Project focusingon the Atlantic spotted dolphins.
The model identifies patterns indolphin vocal sequences and can generate

(06:50):
realistic dolphin like sounds operatingas an audio in audio out system.
It predicts subsequent dolphin soundssimilarly to how language models
predict the next word in human text.
And the AI system is based on Google'slightweight Gemma models and utilizes
sound stream for audio representationWith approximately 400 million parameters,

(07:13):
dolphin Gemma is compact enough to runon pixel phones used in field research.
The advancement could revolutionizeour understanding of dolphin
communication, providing insightsinto their social behaviors, and
potentially enabling interactiveexchanges between humans and dolphins.
By identifying recurring soundpatterns and sequences, researchers

(07:36):
hope to uncover hidden structures andmeanings within dolphin vocalizations.
Google plans to release dolphin Gemmaas an open model this summer, allowing
researchers worldwide to adapt itfor study of other setian species.
Fine tuning will be necessary forspecies with different vocal patterns,
but the open source nature of themodel facilitates adaptations.

(07:59):
So a large language model mayprovide the key to a small
and compact language system.
And if we do find a way to deciphera more complex language, I'm betting
the first decipher sentence willbe, what's taken you so long?
That's our show for today.
I'm your host, Jim Love.

(08:20):
Have a terrific Tuesday.