Sign in

Alexios Mantzarlis

@mantzarlis.com
3.7K followers 1.3K following 2.6K posts

Tech, facts, and carbonara. mantzarlis.com

PostsRepliesMedia
Alexios Mantzarlis @mantzarlis.com · 13h
Overpaying isn’t the only risk facing people who click on ads for such sites. TechCrunch reported earlier this year that one of the sites in my sample “publicly exposed thousands of passports and selfie photos of applicants who paid the site to obtain a U.K. immigration visa.”
techcrunch.com
UK Visa Portal exposed thousands of applicants’ passports and selfies — then called the lawyers on us | TechCrunch
The third-party website exposed passports, selfies, and the location data of applicants who submitted their documents as part of the U.K. visa application process. Instead of fixing the issue, the web...
000
Alexios Mantzarlis @mantzarlis.com · 13h
These sites know what they are doing. Here's one of them that put the domain name for the UK government inside its own ad headline. You can certainly see how people might believe it to be the official website, what with its favorable placement too
100
Alexios Mantzarlis @mantzarlis.com · 13h
These sites would suffer greatly if Google refused their money. Ten of the 16 I reviewed got 50% or more of their traffic from paid search, with six of these relying on it for 75% or more of their visits.
111
Alexios Mantzarlis @mantzarlis.com · 13h
Unofficial providers of visas are flooding Google with thousands of ads and charging 2x to 30x the normal price for travel authorization to Australia, Canada, Egypt, India, Indonesia, New Zealand, Thailand, UK, & US. Only 8% of these ads carry the mandated "Not a government website" disclosure.
121
Alexios Mantzarlis @mantzarlis.com · 13h
Today on @indicator.media: Google is breaking its promise to label ads for unofficial government services providers. Travelers are paying the price.
indicator.media
Google promised to label ads for unofficial visa providers. It’s doing a lousy job of it.
An Indicator audit found labels on only 8% of ads for visa providers that overcharge their users
165
Alexios Mantzarlis @mantzarlis.com · 18h
i had not realized george osborne was working for openai. meta had clegg. pick your fighters, i guess
082
Reposted by Alexios Mantzarlis
Alejandra Caraballo @esqueer.net · 27/09/2026
Google finally updated some of its satellite images of Gaza and my god, it's horrific. The entirety of Rafah has been demolished. The majority of Khan Younis is in ruins. Entire towns completely leveled and erased of the map.
9352173196
Alexios Mantzarlis @mantzarlis.com · 25/09/2026
Thank you!!!
000
Alexios Mantzarlis @mantzarlis.com · 25/09/2026
Great question! Ideally UK, US, Australia, New Zealand, Egypt, Canada, India, Indonesia or Thailand but if it's something else I'm also happy to learn
110
Alexios Mantzarlis @mantzarlis.com · 25/09/2026
i'm looking to chat with folks who overpaid for a tourist visa recently because they landed on the website in a rush searching on Google and it looked like the real deal... ping me here or at alexios at indicator dot media plsss
122
Alexios Mantzarlis @mantzarlis.com · 25/09/2026
Big Friday energy
040
Reposted by Alexios Mantzarlis
Sacha Altay @sachaltay.bsky.social · 22/09/2026
With Claude I did a systematic literature review of misinformation prevalence, exposure, and concentration. There is a companion website that allows you to play with the data and signal me if an estimate is wrong or missing: altay-research.github.io/misinfo-prev...
3138
Alexios Mantzarlis @mantzarlis.com · 23/09/2026
with the little information I see, this new centre sounds a bit like France's VIGINUM
bbc.com
UK to fight Russian disinformation and push new global AI standards, Burnham says
The UK could act as "an honest broker" on AI and other global challenges, the prime minister tells world leaders at the UN.
010
Alexios Mantzarlis @mantzarlis.com · 18/09/2026
burn all the glasses down
futurism.com
Influencers Are Using Meta's AI-Glasses to Film Manipulative Feel-Good Slop of People in Public
A content creator is using facial recognition software paired with Meta's AI glasses to film "joybait" of people without their knowledge.
040
Alexios Mantzarlis @mantzarlis.com · 17/09/2026
❤️ but do you have the agents fight over whether they prefer Craig's articles or mine??
010
Reposted by Alexios Mantzarlis
Renee DiResta @noupside.bsky.social · 17/09/2026
This was a fun convo! It was also sort of an intimidating one since I don’t normally talk ab the process —which is often very hacky and experimental!— behind early investigations. I love Indicator. I regularly forward it to an agent that keeps track of tools & techniques for me. 🔎 Subscribe!
1196
Alexios Mantzarlis @mantzarlis.com · 17/09/2026
Nice deep dive into an operation mentioned in Anthropic's most recent threat report
theverge.com
The sexy AI-powered dating app scams are here
Thousands of people have been catfished by scammers supercharged with AI.
011
Alexios Mantzarlis @mantzarlis.com · 17/09/2026
We continue to track the share of AI contributed posts on X's Community Notes on @indicator.media. Last Friday we saw a new high, with more than two-thirds (67.94%) of the notes coming from a bot. indicator.media/ai-community...
010
Alexios Mantzarlis @mantzarlis.com · 17/09/2026
if you haven't bought Renée's book already...what are you waiting for? It's out now on paperback. I reviewed it back in 2024: indicator.media/p/faked-up-s...
indicator.media
Three books about digital deception
The Death of Truth | Invisible Rulers | The Lie Detectives
010
Alexios Mantzarlis @mantzarlis.com · 17/09/2026
Pod drop alert! The one and only @noupside.bsky.social joined me and @craigsilverman.bsky.social on @indicator.media's Show & Tell to chat Grokipedia, Community Notes, and the wobbly state of platform disclosures on influence operations.
youtube.com
How Renée DiResta realized Grokipedia’s suddenly stopped accepting edits
YouTube video by Indicator
186
Alexios Mantzarlis @mantzarlis.com · 17/09/2026
accountability commenting right here
010
Alexios Mantzarlis @mantzarlis.com · 17/09/2026
IDK if I'm going to take this Masterclass from you, Pedro
120
Alexios Mantzarlis @mantzarlis.com · 15/09/2026
This is the 116th reported case logged on @indicator.media's map of AI nudifiers being used in school communities, launched with @mattburgess1.bsky.social or @wired.com earlier this year. indicator.media/nudifiers-un...
032
Alexios Mantzarlis @mantzarlis.com · 15/09/2026
Father files police report against son after he creates fake nude photos of a classmate using AI.
metropoles.com
Pai registra BO contra filho após criação de nudes falsos de colega com IA
Adolescente, de 15 anos, teria contribuído na criação e no compartilhamento de imagens íntimas falsas com o rosto de uma colega da escola
120
Reposted by Alexios Mantzarlis
Brent Toderian @brenttoderian.bsky.social · 12/09/2026
If walking costs you $1, we all pay $0.01. If biking costs you $1, we all pay $0.08. If bussing costs you $1, we all pay $1.50. If driving costs you $1, we all pay $9.20. Via study that still underestimates climate cost. This isn’t about choice. It’s about who pays for your choice. #citymakingmath
Detailed math as shared in the post. “How much does your commute cost (or save) society? Source is Discourse Media.
5393173
Alexios Mantzarlis @mantzarlis.com · 10/09/2026
oh no, the US government actually asked decent questions about content moderation and online information quality (this re: the UK's green paper on boosting "trustworthy media providers" online) uk.usembassy.gov/u-s-governme...
010
Alexios Mantzarlis @mantzarlis.com · 10/09/2026
Meta does not publish downloadable data about CN like X does, so guerrilla audits like this one are all we can do. Paid members make Indicator's work possible -- and also get access to the full dataset underpinning this analysis. Share, like, subscribe ---->
indicator.media
Meta says Community Notes is bigger than fact-checking. That's not the whole story
Crowdsourced fact-checking on Instagram covered just 3% of false posts debunked by one fact-checking website
020
Alexios Mantzarlis @mantzarlis.com · 10/09/2026
Meta says that it's still early days. But it sure looks like Community Notes should complement rather than substitute fact-checkers. That they have been presented as alternatives is entirely down to Mark Zuckerberg's desire to please the Trump administration.
120
Alexios Mantzarlis @mantzarlis.com · 10/09/2026
Posts that have gone without a note despite AFP publishing a fact check about them include false claims about children’s vaccines, the health benefits of nicotine pouches, and voter fraud in the NYC mayoral election. Identical posts on X and Instagram carry a note on the first but not the latter.
140
Alexios Mantzarlis @mantzarlis.com · 10/09/2026
Maybe. But Community Notes appear to be targeting something different from fact-checkers. There are only 5 Community Notes on the 166 Insta posts fact-checked by AFP's US bureau since April 2025. This is much lower than the 20.5% rate on X, hardly an oasis of quality information.
131
Alexios Mantzarlis @mantzarlis.com · 10/09/2026
Meta said yesterday that it will expand crowdsourced fact-checking to 16 Latin American countries, claiming that “the daily volume of Community Notes has doubled independent fact-checks" compared to its defunct US fact-checking program.
about.fb.com
Iniciamos los Tests de Notas de la Comunidad en Hispanoamérica | Acerca de Meta
En enero del 2025, lanzamos las Notas de la Comunidad en Estados Unidos: una forma colaborativa para que las personas que utilizan nuestras plataformas aporten información contextual a publicaciones e...
120
Alexios Mantzarlis @mantzarlis.com · 10/09/2026
New on @indicator.media: Community Notes on Instagram covered just 3% of false posts debunked by the US bureau of AFP Fact Check. This analysis comes one day after Meta announced it would expand the program to 16 new countries.
indicator.media
Meta says Community Notes is bigger than fact-checking. That's not the whole story
Crowdsourced fact-checking on Instagram covered just 3% of false posts debunked by one fact-checking website
1169
Alexios Mantzarlis @mantzarlis.com · 09/09/2026
Meta is expanding Community Notes to 16 Latin American countries, data still nowhere to be seen.* *I've got a story coming out tomorrow about this, as chance will have it
about.fb.com
Iniciamos los Tests de Notas de la Comunidad en Hispanoamérica | Acerca de Meta
En enero del 2025, lanzamos las Notas de la Comunidad en Estados Unidos: una forma colaborativa para que las personas que utilizan nuestras plataformas aporten información contextual a publicaciones e...
020
Alexios Mantzarlis @mantzarlis.com · 03/09/2026
And the kicker is that about half of them are for repeat offenders, actors known to Meta that are barely concealing their activity. I'm sick and tired of doing this kind of audit but I also know that unless the drumbeat goes on, the incentives to do better fade away.
indicator.media
Meta ran at least 11,000 ads for AI nudifiers this year – many from easily detectable repeat offenders
Indicator’s latest audit reveals the company let a network of identifiable bad actors run thousands of ads for nonconsensual deepfake nudes
000
Alexios Mantzarlis @mantzarlis.com · 03/09/2026
And yet here I am with my little burner accounts and Claude Max subscription still finding 11,000 ads for SYNCII-generators on Facebook and Instagram, 14% of which Meta never actioned.
indicator.media
Meta ran at least 11,000 ads for AI nudifiers this year – many from easily detectable repeat offenders
Indicator’s latest audit reveals the company let a network of identifiable bad actors run thousands of ads for nonconsensual deepfake nudes
100
Alexios Mantzarlis @mantzarlis.com · 03/09/2026
In March of this year, the company said that with AI it "can catch more of these violations more accurately."
about.fb.com
Boosting Your Support and Safety on Meta's Apps With AI
We’re launching new AI tools for support and content enforcement on our apps to make them work better for you.
100
Alexios Mantzarlis @mantzarlis.com · 03/09/2026
In June 2025, Meta said it was “building new technology to detect ads for nudify apps.”
about.fb.com
Combating Nudify Apps with Lawsuit & New Technology | Meta
Meta is committed to protecting our users from nudify apps, which is why we're taking action with new technology and legal measures.
110
Alexios Mantzarlis @mantzarlis.com · 03/09/2026
In his infamous moderation pivot of January 2025, Mark Zuckerberg promised to focus proactive enforcement activity on “illegal and high-severity violations.”
about.fb.com
More Speech and Fewer Mistakes
We're ending our third party fact checking program and moving to a Community Notes model.
120
Alexios Mantzarlis @mantzarlis.com · 03/09/2026
New on @indicator.media: Simple text searches for repeat offenders could have caught ads for AI nudifiers on Meta and prevented them from being served to millions of Meta users this year.
indicator.media
Meta ran at least 11,000 ads for AI nudifiers this year – many from easily detectable repeat offenders
Indicator’s latest audit reveals the company let a network of identifiable bad actors run thousands of ads for nonconsensual deepfake nudes
11911
Reposted by Alexios Mantzarlis
joe bodnar @joeabodnar.bsky.social · 02/09/2026
Amazing that this renewed interested in foreign disinfo is being driven by overhyped reports about Chinese interference in data center debates. The X update mirrored OpenAI's report, which categorized the Chinese operation (Spamo) as Category One on the breakout scale. That's the lowest category.
131
Alexios Mantzarlis @mantzarlis.com · 02/09/2026
Once again: The Trumpian crusade against Trust & Safety was never about free speech. It was always about making sure platform policies applied only to adversaries and never to those shouting "censorship" for facing consequences after denying election results or the existence of COVID-19.
reneediresta.com
Suddenly, Foreign Influence Research is Good Again
What X’s Chinese bot takedown tells us about the counter-disinformation infrastructure we dismantled   On Monday I wrote about Under Secretary for Public Diplomacy Sarah Rogers’ extraordinary effor...
042
Alexios Mantzarlis @mantzarlis.com · 27/08/2026
ref.
newyorker.com
How CoComelon Captures Our Children’s Attention
The animation juggernaut is now streamed for billions of hours each year, including on Netflix and its own YouTube channel. Should we be worried about that?
000
Alexios Mantzarlis @mantzarlis.com · 27/08/2026
on the one hand: bad on the other hand: this is already a terrible show
404media.co
Cocomelon's Studio Tells Its Artists to Start Experimenting With AI
Moonbug Entertainment, which makes Cocomelon, Blippi, and other popular content for children, says it will begin using AI but will always have a 'human in the loop.'
120
Alexios Mantzarlis @mantzarlis.com · 27/08/2026
The important thing that isn't clear is whether this team rolls into the VP of Trust & Safety or into the VP of Public Policy. If it goes into T&S this is a reasonable move; much less so the latter.
wsj.com
Exclusive | Google Moves AI-Responsibility Team Out of DeepMind Lab in Latest Shake-Up
Researchers have raised concerns about the change’s effect on their independence and ability to detect threats from newest models.
041
Alexios Mantzarlis @mantzarlis.com · 26/08/2026
ok since no one asked www.youtube.com/shorts/9FoZh...
youtube.com
This fake employee scam got 100 million views on Facebook
YouTube video by Indicator
000
Alexios Mantzarlis @mantzarlis.com · 26/08/2026
it took me 90 minutes to get a terrible 90 second short video out there i have entirely new respect for youtubers
130
Alexios Mantzarlis @mantzarlis.com · 26/08/2026
Also as self-serving as this comment may have been, I'm glad to take it as a sign that Trust & Safety ain't going anywhere
030
Alexios Mantzarlis @mantzarlis.com · 26/08/2026
We tracked it back to a single scammy affiliate marketing operation which pushed almost 8,000 reels with slightly altered assets to evade moderation. This story led Meta to remove all 107 accounts and Lovable to shut down ten domains. Subscribe to @indicator.media to support our impact.
indicator.media
AI-generated Costco employees got over 100 million views on Facebook and Instagram
We tracked it back to a single scammy affiliate marketing operation
162
Alexios Mantzarlis @mantzarlis.com · 26/08/2026
Today on Indicator: AI-generated fake employees got over 100 million views on Facebook and Instagram by pushing fake gift cards for ALDI, Costco, The Home Depot, Target, and more.
indicator.media
AI-generated Costco employees got over 100 million views on Facebook and Instagram
We tracked it back to a single scammy affiliate marketing operation
2179
Alexios Mantzarlis @mantzarlis.com · 25/08/2026
“Find out who you are and do it on purpose” RIP Dolly www.nytimes.com/2026/08/25/a...
nytimes.com
Dolly Parton, Country Music Superstar Whose Songs Spoke to Millions, Dies at 80 (Gift Article)
Known for her outsize persona, appearance and talent, she rose from rural poverty to become a top-selling singer, songwriter and entrepreneur.
061