Digimasters Shorts - Delphi-2M AI Predicts Disease Risks, Google Gemini Goes Free in Chrome, OpenAI Deep Research Flaw Exposed, Zoom Unveils AI Companion 3.0, OpenAI Teams with Jony Ive for Secret AI Devices

Send us a text

Digimasters Shorts keeps you up-to-date with the latest in the digital world. Join hosts Adam Nagus and Carly Wilson as they explore breakthrough AI innovations, including predictive health models, expanded Google AI tools, and emerging security challenges in AI deployment. Discover how tech giants like Google and OpenAI are shaping the future with new features, security alerts, and even upcoming consumer devices. Stay informed about the evolving landscape of digital security, productivity tools, and cutting-edge developments in AI hardware—all in quick, insightful episodes designed to keep you ahead in the age of technology.

Support the show

Don't forget to checkout our larger sister podcast - The Digimasters Podcast here. Which has many expert guests discussing AI, Career Mentoring, Fractional Careers, Digital and much much more.

Episode Transcript

Adam N2 (00:05):
Welcome to Digimasters Shorts, we are your hosts Adam
Nagus

Carly W (00:09):
and Carly Wilson delivering the latest scoop from
the digital realm.
Scientists have developed anartificial intelligence model
capable of predicting people'srisk of over 1,000 diseases up
to a decade in advance byanalyzing patterns in medical
records.
This technology functions like aweather forecast for health,
estimating probabilities ratherthan exact dates for conditions

(00:29):
such as type 2 diabetes, heartattacks, and sepsis.
The AI, named Delphi-2M, wastrained on anonymous U.K medical
data from over 400,000individuals and validated with
records from nearly 2 millionpeople in Denmark.
Researchers believe this toolcould identify high-risk
patients early, potentiallyguiding preventative measures
like medication or lifestylechanges.

(00:52):
It may also help hospitalsanticipate local healthcare
demands years ahead, improvingresource planning.
Although promising, the AI isnot yet ready for clinical use
and requires further testing andrefinement, including
adjustments to addressdemographic biases.
Experts caution that thetechnology needs robust
regulation beforeimplementation, comparing its
potential impact to the rise ofgenomics in medicine.

(01:15):
The study, published in thejournal Nature, is a
collaboration between Europeanand Danish research
institutions.
AI researchers emphasize theimportance of ethical
development to ensureresponsible deployment in
healthcare.
Overall, this advancement marksa significant step toward
personalized and predictivemedicine at scale.

Adam N2 (01:34):
Google is expanding its Gemini AI assistant in Chrome to
all Mac and Windows desktopusers in the U.S., having
previously limited it to paidsubscribers.
U.S.
users with English languagesettings can now use Gemini to
clarify complex information onany web page through an icon in
the browser.
Gemini supports multitaskingacross tabs, allowing users to

(01:54):
compare and summarizeinformation, such as trip
planning or shopping.
Upcoming features include theability to recall previously
visited web pages, making iteasier to revisit past browsing
sessions.
Integration with other Googleapps like Calendar, YouTube, and
Maps will let users schedulemeetings and access location
details without leaving theircurrent page.

(02:15):
Gemini will also handle taskslike booking haircuts or
ordering groceries, navigatingsites and adding items to carts
for users to finalize.
Google plans to introduce AIMode search in the Chrome
address bar, enabling complexqueries and follow-ups directly
from there.
This feature will roll out inthe U.S.
later this month and expand toother countries and languages.

(02:36):
Chrome will use the Gemini Nanomodel to detect and block
AI-generated scams such asphishing alerts.
Additionally, Chrome will offerautomatic password resets on
supported sites if a breach isdetected, enhancing user
security.
Security researchers haveuncovered and plugged a
vulnerability in Open A.I's DeepResearch AI tool embedded within

(02:56):
Chat G.P.T.
This flaw, dubbed Shadow Leak,exploited a prompt injection
attack to co-opt the AI intostealing sensitive data from
Gmail inboxes without userdetection.
The attack relied on AI agents’ability to operate autonomously
after being granted access topersonal emails and documents.
Radware, the security firmbehind the research,

(03:17):
demonstrated how hiddeninstructions embedded in emails
could trigger the AI to searchfor and export confidential
information.
This method ran undetected onOpen A.I’s cloud infrastructure,
making it invisible to standardcybersecurity defenses.
The researchers emphasized thedifficulty of pulling off such
an exploit, describing a processfilled with trial and error

(03:37):
before success.
Shadow Leak also highlights thebroader risks associated with
outsourcing tasks to AI agentsacross platforms like Outlook,
GitHub, Google Drive, andDropbox.
Radware warned that similarattacks could expose sensitive
business data, includingcontracts and meeting notes.
The study serves as a crucialwarning about the emerging
security challenges posed byagentic AI technologies.

(04:01):
Users and companies must stayvigilant as AI integration
deepens into daily workflows.

Carly W (04:06):
Zoom has unveiled AI Companion 3.0 at its Zoomtopia
conference, aiming to boostworker productivity and compete
with other video communicationplatforms.
This upgraded tool uses agenticAI to analyze meetings, calls,
chat history, and documents toprovide timely, critical
information.
Enterprise Zoom Workplace userscan now enhance and organize

(04:28):
their notes across multipleplatforms, including Microsoft
Teams and Google Meet.
The AI also offers"outcome-focused prompts" to
help manage administrativetasks, suggesting meetings to
skip and dedicated work times,though human approval is
required for changes.
A new"work surface" featureassists with creating detailed
reports and polished documentsby synthesizing scattered

(04:50):
information.
Zoom Workplace will also gainlifelike avatars, video clip
generation, and real-time voicetranslation for multilingual
meetings.
The avatars can mimic meetingactions for users who prefer not
to appear on camera.
AI Companion 3.0 will beavailable starting November,
with optional paid add-ons forcustom AI agents.

(05:12):
While AI productivity toolsgrow, concerns about privacy and
job displacement remainprevalent.
Research indicates workerproductivity is rising, but
wages have not kept pace withthese gains.
Open A.I is reportedlydeveloping a line of AI devices
in collaboration with formerApple design chief Jony Ive.
The first device is said toresemble a smart speaker without

(05:33):
a display.
The company has securedcontracts with Apple assemblers
Luxshare and Goertek to supplycomponents like speaker modules.
Other potential devices includeAI glasses, a digital voice
recorder, and a wearable pin,with releases expected in late
2026 or early 2027.
C.E.O Sam Altman described theinitial product as pocket-sized,

(05:56):
contextually aware, andscreen-free.
While Altman ruled out glassesas the first product, smart
eyewear may still come later.
Reports suggest Apple canceled ameeting to prevent executive
defections to Open A.I.
Several Apple hardware employeeshave joined Open A.I, attracted
by promises of less bureaucracyand more collaboration.
Open A.I’s chief hardwareofficer Tang Tan, formerly of

(06:19):
Apple, is leading recruitmentefforts.
The partnership marks asignificant push by Open A.I to
enter the consumer hardwaremarket using Apple’s supply
chain expertise.

Don (06:29):
Thank you for listening to today's AI and Tech News podcast
summary...
Please do leave us a comment andfor additional feedback, please
email us [e-mail] You
can now follow us on Instagramand Threads by searching for
@DigimastersShorts or Search forDigimasters on Linkedin.
Be sure to tune in tomorrow anddon't forget to follow or
subscribe!