Alexios Mantzarlis
banner
mantzarlis.com
Alexios Mantzarlis
@mantzarlis.com
Tech, facts, and carbonara.

mantzarlis.com
Pinned
NEW on @indicator.media: A first *full-scale* comparison of Grokipedia v Wikipedia.

Last week the awesome @harold.bsky.social rocked up to my desk bearing gifts.

Hal had collected almost all 900K Grokipedia entries and compared them to their Wikipedia equivalents for text and citation similarity.
Grokipedia cites a Nazi forum and fringe conspiracy websites
A site-wide comparison with Wikipedia sheds light on what Elon Musk is trying to do
indicator.media
Reposted by Alexios Mantzarlis
Would love to think it was a passive-aggressive swipe at a conservative structure which never really embraced him. Or maybe a wayward employee who snuck one past the boss.
I'd love to know why Musk thought this was a good idea. I mean, it was for us, not so much for him.
November 24, 2025 at 4:50 PM
Reposted by Alexios Mantzarlis
I wrote about the fake account blowup on X this weekend. A genuine post-truth nightmare and proof that these companies have polluted their platforms so thoroughly and traded reality for profit that they've undermined the very idea of what the internet is supposed to be.
That MAGA Account Might Be a Troll From Pakistan
How X blew up its own platform with a new location feature
www.theatlantic.com
November 24, 2025 at 4:33 PM
A major perk for @indicator.media members is our monthly workshop. This Friday, @craigsilverman.bsky.social will walk folks through how to look for valuable documents hiding in open air, the topic of our latest guide. Upgrade to paid now to attend the workshop!
The Indicator Guide to hunting for documents and files in open buckets, servers, and directories
How to find interesting and potentially confidential documents sitting on a publicly-accessible server or website
indicator.media
November 24, 2025 at 12:47 PM
MAXIIIIIIIIIII
November 24, 2025 at 1:24 AM
There really is only one Rose Lavelle
The Bruninha stepover ✅
The Rose Lavelle rocket 🚀
The Gotham celly 🔥

#NWSLChampionship | Google Pixel
November 23, 2025 at 3:12 PM
Reposted by Alexios Mantzarlis
Remember all those promises to Cassidy about vaccines??
CDC has overhauled its website to assert that “the claim ‘vaccines do not cause autism’ is not an evidence-based claim”
November 20, 2025 at 2:43 AM
Reposted by Alexios Mantzarlis
He just tweeted it out
An X user asks Elon Musk why they’re seeing tweets from left-wing lawmakers.

Musk: “Because we are failing very badly with the recommendations algorithm. Doing my best to address this.”
November 21, 2025 at 12:50 PM
Today on @indicator.media's (free) Briefing: A twinkle of optimism on the AI slop front as TikTok and Google roll out tools to filter and detect AI content.

I remain bearish about generative AI and info quality. But I’ll celebrate developments that increase user agency and platform transparency.
Briefing: Good news on the AI slop front. No, really!
Plus: detecting AI-generated images with SynthID, and advanced archival OSINT
indicator.media
November 21, 2025 at 12:54 PM
Reposted by Alexios Mantzarlis
Over on X today, Elon Musk's AI chatbot Grok is having another normal one www.washingtonpost.com/technology/2...
Elon Musk’s Grok chatbot ranks him as world history’s greatest human
Users on X shared examples of the “truth-seeking” AI chatbot praising its owner as “strikingly handsome,” a “genius,” and fitter than LeBron James.
www.washingtonpost.com
November 20, 2025 at 11:55 PM
Reposted by Alexios Mantzarlis
why is every day in 2025 a worse and worse game of madlibs?
November 20, 2025 at 10:41 PM
Reposted by Alexios Mantzarlis
NEW: Elon Musk's Grokipedia cites a neo-Nazi website 42 times, researchers say www.nbcnews.com/tech/elon-mu...
Elon Musk’s Grokipedia cites Stormfront — a neo-Nazi forum — dozens of times, study finds
An analysis by researchers at Cornell University is the first comprehensive look at Grokipedia since Musk launched his project last month.
www.nbcnews.com
November 20, 2025 at 3:22 PM
Reposted by Alexios Mantzarlis
TikTok & Instagram have been deluged by thousands of videos from female students that praise Jenni, an AI study app. They say it's better than ChatGPT. They say it can improve your grades.

What don't they say? That they were paid to promote Jenni.
Thousands of undisclosed ads flooded TikTok to sell AI study apps
Maryland-based Mindgrasp used an army of student creators (and actors) to promote its study tools, without disclosing that they were paid.
indicator.media
November 20, 2025 at 1:58 PM
Reposted by Alexios Mantzarlis
Trump says he’s meeting with “communist” Mamdani at the White House on Friday
November 20, 2025 at 1:32 AM
Reposted by Alexios Mantzarlis
@davidingram.bsky.social covered @mantzarlis.com and my work on Grokipedia citations for NBC! much more to analyze here

www.nbcnews.com/news/amp/rcn...
Elon Musk’s Grokipedia cites neo-Nazi website 42 times: study
An analysis by researchers at Cornell University is the first comprehensive look at Grokipedia since Musk launched his project last month.
www.nbcnews.com
November 20, 2025 at 1:26 PM
"The current CDC page says that studies supporting a link between vaccines and autism 'have been ignored by health authorities.'"
Under RFK Jr., CDC promotes false vaccines-autism link it once discredited
The CDC’s website now says health authorities ignored evidence of a potential connection between vaccines and autism, despite dozens of studies showing no link.
www.washingtonpost.com
November 20, 2025 at 1:20 PM
Reposted by Alexios Mantzarlis
Just wanna re-up in simple terms that when Biden talked to platforms, Jim Jordan launched years of investigations into everybody involved, said it was tyranny, censorship, etc.

And now they just straight up acknowledge that they talk to platforms too.

www.washingtonexaminer.com/news/crime/3...
DHS playing 'whack-a-mole' shooting down made-up ICE stories
The Department of Homeland Security is stepping up efforts to combat fake news, viral AI videos, and misinformation on ICE and Border Patrol.
www.washingtonexaminer.com
November 20, 2025 at 1:54 AM
Reposted by Alexios Mantzarlis
Now that is an interesting finding.
The AI bots are getting their notes approved at double the rate of humans.

Should X go from 8 AI contributors to 80, this could dramatically upend the composition and purpose of the program.
November 19, 2025 at 7:31 PM
Reposted by Alexios Mantzarlis
This is...sort of brilliant.
November 19, 2025 at 12:56 PM
Reposted by Alexios Mantzarlis
I think the word you’re looking for is “kept”
November 19, 2025 at 2:14 AM
Reposted by Alexios Mantzarlis
we also open source all of our code, data, and embeddings!
paper: arxiv.org/abs/2511.09685
github: github.com/htried/wiki-...
huggingface: huggingface.co/datasets/htr...
November 17, 2025 at 4:11 PM
New on @indicator.media: The single most active contributor to Community Notes since Sep. 1 is Zesty Walnut Grackle.

Zesty is an AI bot, one of eight currently active on X after the platform onboarded them in a bid to boost flagging participation.
An AI bot is now the top contributor to Community Notes on X
But synthetic contributors aren’t replacing humans – yet
indicator.media
November 18, 2025 at 2:18 PM
Reposted by Alexios Mantzarlis
Researchers tried plugging every possible phone number into WhatsApp's web app. They found they could collect 3.5 billion users' phone numbers, plus photos for half and profile text for more than a third, the biggest personal data exposure ever by some measures. www.wired.com/story/a-simp...
A Simple WhatsApp Security Flaw Exposed 3.5 Billion Phone Numbers
By plugging tens of billions of phone numbers into WhatsApp’s contact discovery tool, researchers found “the most extensive exposure of phone numbers” ever—along with profile photos and more.
www.wired.com
November 18, 2025 at 2:04 PM