Jeff Horwitz
@jeffhorwitz.bsky.social
Reporter covering Meta, social media and other topics (hopefully) for Reuters' investigative Team.
Reposted by Jeff Horwitz
AI porn generator and ERP chatbot Secret Desires left millions of images and videos exposed in unsecured cloud storage. I took a look and found folders called "faceswap," where people had uploaded tons of stolen content and random photos of women www.404media.co/ai-porn-secr...
Massive Leak Shows Erotic Chatbot Users Turned Women’s Yearbook Pictures Into AI Porn
Chatbot roleplay and image generator platform SecretDesires.ai left cloud storage containers of nearly two million of images and videos exposed, including photos and full names of women from social me...
www.404media.co
November 19, 2025 at 3:25 PM
AI porn generator and ERP chatbot Secret Desires left millions of images and videos exposed in unsecured cloud storage. I took a look and found folders called "faceswap," where people had uploaded tons of stolen content and random photos of women www.404media.co/ai-porn-secr...
For the first time ever, I had to make a payment in crypto.
I can now confidently say that crypto payments combine the speed of a dial up modem and the ease of updating the drivers on a 1998 HP Laserjet printer.
I can now confidently say that crypto payments combine the speed of a dial up modem and the ease of updating the drivers on a 1998 HP Laserjet printer.
November 18, 2025 at 5:58 AM
For the first time ever, I had to make a payment in crypto.
I can now confidently say that crypto payments combine the speed of a dial up modem and the ease of updating the drivers on a 1998 HP Laserjet printer.
I can now confidently say that crypto payments combine the speed of a dial up modem and the ease of updating the drivers on a 1998 HP Laserjet printer.
Mystery Fuels Unease in Maine Woods: Who Bought Burnt Jacket Mountain?
www.nytimes.com
November 16, 2025 at 5:33 PM
I knew doing scam research would put me into the "dumbest guy alive" digital ad targeting category, but nonexistent Milwaukee Brand cyberpunk staplers?
Holy hell I don't think I was ready for the ad-based internet to get even worse.
Holy hell I don't think I was ready for the ad-based internet to get even worse.
November 14, 2025 at 6:08 PM
I knew doing scam research would put me into the "dumbest guy alive" digital ad targeting category, but nonexistent Milwaukee Brand cyberpunk staplers?
Holy hell I don't think I was ready for the ad-based internet to get even worse.
Holy hell I don't think I was ready for the ad-based internet to get even worse.
Wrote about this a few weeks ago, but have an update:
X.com continues to stand up to the legal demands of British Columbia, which has demanded that the platform remove... non consensual intimate imagery in response to a court order.
news.gov.bc.ca/releases/202...
X.com continues to stand up to the legal demands of British Columbia, which has demanded that the platform remove... non consensual intimate imagery in response to a court order.
news.gov.bc.ca/releases/202...
Attorney general’s statement about social media company X challenging Civil Resolution Tribunal order
Niki Sharma, Attorney General, has released the following statement in response to social media company X applying for judicial review of a Civil Resolution Tribunal (CRT) order to remove a non-consensual intimate image from its platform:
news.gov.bc.ca
November 14, 2025 at 5:58 PM
Wrote about this a few weeks ago, but have an update:
X.com continues to stand up to the legal demands of British Columbia, which has demanded that the platform remove... non consensual intimate imagery in response to a court order.
news.gov.bc.ca/releases/202...
X.com continues to stand up to the legal demands of British Columbia, which has demanded that the platform remove... non consensual intimate imagery in response to a court order.
news.gov.bc.ca/releases/202...
Reposted by Jeff Horwitz
November 8, 2025 at 5:29 PM
Meant to post this from one of underlying documents:
Meta earns $3.5B ever six months in "higher legal risk" revenue from scams, and that such amount is could be the cumulative "outside order of magnitude" for fines from regulators for accepting scam ads.
www.reuters.com/investigatio...
Meta earns $3.5B ever six months in "higher legal risk" revenue from scams, and that such amount is could be the cumulative "outside order of magnitude" for fines from regulators for accepting scam ads.
www.reuters.com/investigatio...
November 6, 2025 at 8:58 PM
Meant to post this from one of underlying documents:
Meta earns $3.5B ever six months in "higher legal risk" revenue from scams, and that such amount is could be the cumulative "outside order of magnitude" for fines from regulators for accepting scam ads.
www.reuters.com/investigatio...
Meta earns $3.5B ever six months in "higher legal risk" revenue from scams, and that such amount is could be the cumulative "outside order of magnitude" for fines from regulators for accepting scam ads.
www.reuters.com/investigatio...
This is terrible news, as is the original decision to fold Teen Vogue -- it consistently did some of the best culture reporting on social media of any outlet, period.
Great respect to those who stood up on behalf of their newsrooms.
Great respect to those who stood up on behalf of their newsrooms.
New: Conde Nast fired four employees who were among a group that confronted the company's head of human resources on Wednesday over the decision to fold Teen Vogue into Vogue/recent cuts. Employees who were fired included journalists from the New Yorker, Wired, and Bon Appétit.
November 6, 2025 at 6:05 PM
This is terrible news, as is the original decision to fold Teen Vogue -- it consistently did some of the best culture reporting on social media of any outlet, period.
Great respect to those who stood up on behalf of their newsrooms.
Great respect to those who stood up on behalf of their newsrooms.
Meta earns $3.5 billion every six months from showing Faceboon and Instagram users 15 billion “higher legal risk” scam ad impressions a day, internal documents state.
That haul vastly exceeds how much the company expects regulators
To fine it for running scam ads.
www.reuters.com/investigatio...
That haul vastly exceeds how much the company expects regulators
To fine it for running scam ads.
www.reuters.com/investigatio...
www.reuters.com
November 6, 2025 at 11:46 AM
Meta earns $3.5 billion every six months from showing Faceboon and Instagram users 15 billion “higher legal risk” scam ad impressions a day, internal documents state.
That haul vastly exceeds how much the company expects regulators
To fine it for running scam ads.
www.reuters.com/investigatio...
That haul vastly exceeds how much the company expects regulators
To fine it for running scam ads.
www.reuters.com/investigatio...
FACEBOOK MARKETPLACE AND MY FEET: A MYSTERY
My wife wanted to sell a mirror, so I listed it on Facebook Marketplace this weekend. I neglected to wear shoes for the photo.
I have since received seven offers: two for the mirror, and five for acts involving my feet. I'm now curious what's going on.
My wife wanted to sell a mirror, so I listed it on Facebook Marketplace this weekend. I neglected to wear shoes for the photo.
I have since received seven offers: two for the mirror, and five for acts involving my feet. I'm now curious what's going on.
October 29, 2025 at 1:23 AM
FACEBOOK MARKETPLACE AND MY FEET: A MYSTERY
My wife wanted to sell a mirror, so I listed it on Facebook Marketplace this weekend. I neglected to wear shoes for the photo.
I have since received seven offers: two for the mirror, and five for acts involving my feet. I'm now curious what's going on.
My wife wanted to sell a mirror, so I listed it on Facebook Marketplace this weekend. I neglected to wear shoes for the photo.
I have since received seven offers: two for the mirror, and five for acts involving my feet. I'm now curious what's going on.
I'm skeptical that Twitter/X is going to heavily discount machine learning-based AI engagement farming in favor of generative AI-based curation, but you know why not? Give it a go.
www.theverge.com/news/802480/...
www.theverge.com/news/802480/...
X is changing how it handles links to try and keep you in the app
X is making changes to better serve link-based posts and keep you in the app.
www.theverge.com
October 26, 2025 at 7:37 PM
I'm skeptical that Twitter/X is going to heavily discount machine learning-based AI engagement farming in favor of generative AI-based curation, but you know why not? Give it a go.
www.theverge.com/news/802480/...
www.theverge.com/news/802480/...
I would gladly read a followup on this focused entirely on IRBs. I've been surprised by how often researchers say they could never got approval to run tests mimicking the sort of interventions that major platforms undertake on far larger populations as a matter of course on a daily basis.
October 24, 2025 at 4:42 PM
I would gladly read a followup on this focused entirely on IRBs. I've been surprised by how often researchers say they could never got approval to run tests mimicking the sort of interventions that major platforms undertake on far larger populations as a matter of course on a daily basis.
This recent preprint on tech industry influence over external research is compelling. Beyond documenting the low rate of industry-funded researchers actually disclosing that fact, the paper points out that journals have routinely waived IRB review for industry work.
arxiv.org/abs/2510.19894
arxiv.org/abs/2510.19894
October 24, 2025 at 4:08 PM
This recent preprint on tech industry influence over external research is compelling. Beyond documenting the low rate of industry-funded researchers actually disclosing that fact, the paper points out that journals have routinely waived IRB review for industry work.
arxiv.org/abs/2510.19894
arxiv.org/abs/2510.19894
Reposted by Jeff Horwitz
1. We ( @jbakcoleman.bsky.social, @cailinmeister.bsky.social, @jevinwest.bsky.social, and I) have a new preprint up on the arXiv.
There we explore how social media companies and other online information technology firms are able to manipulate scientific research about the effects of their products.
There we explore how social media companies and other online information technology firms are able to manipulate scientific research about the effects of their products.
October 24, 2025 at 12:47 AM
1. We ( @jbakcoleman.bsky.social, @cailinmeister.bsky.social, @jevinwest.bsky.social, and I) have a new preprint up on the arXiv.
There we explore how social media companies and other online information technology firms are able to manipulate scientific research about the effects of their products.
There we explore how social media companies and other online information technology firms are able to manipulate scientific research about the effects of their products.
Reposted by Jeff Horwitz
This new reporting from @jeffhorwitz.bsky.social is jaw-dropping and I think warrants a response from @cos.io given their ongoing collaboration with Meta.
Jeff writes about how Meta identified content-specific harms as a key problem for teens on Instagram. 🧵
www.reuters.com/business/ins...
Jeff writes about how Meta identified content-specific harms as a key problem for teens on Instagram. 🧵
www.reuters.com/business/ins...
Exclusive: Instagram shows more ‘eating disorder adjacent’ content to vulnerable teens, internal Meta research shows
Meta researchers found that teens who report that Instagram regularly made them feel bad about their bodies saw significantly more “eating disorder adjacent content” than those who did not, according to an internal document reviewed by Reuters.
www.reuters.com
October 20, 2025 at 1:23 PM
This new reporting from @jeffhorwitz.bsky.social is jaw-dropping and I think warrants a response from @cos.io given their ongoing collaboration with Meta.
Jeff writes about how Meta identified content-specific harms as a key problem for teens on Instagram. 🧵
www.reuters.com/business/ins...
Jeff writes about how Meta identified content-specific harms as a key problem for teens on Instagram. 🧵
www.reuters.com/business/ins...
Four years ago, pre GenAI, someone showed me some stunning photos of ravens they'd taken. Among the most beautiful wildlife portraits I've seen. And they refused to put them even on a website because they didn't "want a tech company to steal them."
I rolled my eyes. I think about that now and then.
I rolled my eyes. I think about that now and then.
October 13, 2025 at 7:50 PM
Four years ago, pre GenAI, someone showed me some stunning photos of ravens they'd taken. Among the most beautiful wildlife portraits I've seen. And they refused to put them even on a website because they didn't "want a tech company to steal them."
I rolled my eyes. I think about that now and then.
I rolled my eyes. I think about that now and then.
Reposted by Jeff Horwitz
Here is a reporter who deserves to be remembered.
kyivindependent.com/ukrainian-jo...
kyivindependent.com/ukrainian-jo...
Ukrainian journalist Roshchyna posthumously named Press Freedom Hero
The IPI said Roshchyna's death on Sept. 19, 2024 "exposed the brutal treatment suffered by the dozens of Ukrainian journalists unjustly held by Russian authorities, and brought new urgency to the need...
kyivindependent.com
October 11, 2025 at 3:15 PM
Here is a reporter who deserves to be remembered.
kyivindependent.com/ukrainian-jo...
kyivindependent.com/ukrainian-jo...
Happy birthday to this trade publication, which I am fairly confident will outlive most of its news industry peers
October 7, 2025 at 10:21 PM
Happy birthday to this trade publication, which I am fairly confident will outlive most of its news industry peers
This is really good. Actually testing the models remains underutilized as a reporting methodology.
www.technologyreview.com/2025/10/01/1...
www.technologyreview.com/2025/10/01/1...
www.technologyreview.com
October 6, 2025 at 2:41 AM
This is really good. Actually testing the models remains underutilized as a reporting methodology.
www.technologyreview.com/2025/10/01/1...
www.technologyreview.com/2025/10/01/1...
1. Great story by my former colleagues
2. I don't think copyright law works like that
www.wsj.com/tech/ai/open...
2. I don't think copyright law works like that
www.wsj.com/tech/ai/open...
Exclusive | OpenAI’s New Sora Video Generator to Require Copyright Holders to Opt Out
Executives at the company notified talent agencies and studios over the last week.
www.wsj.com
September 30, 2025 at 12:29 AM
1. Great story by my former colleagues
2. I don't think copyright law works like that
www.wsj.com/tech/ai/open...
2. I don't think copyright law works like that
www.wsj.com/tech/ai/open...
Reposted by Jeff Horwitz
In case anyone thought ICE shooting at journalists was just an LA thing. Or that they'd cut it out after repeated warnings from judges. Nope. More of the same in Chicago.
Took fire to the face. ICE definitely isn’t aiming at ground and firing at people’s heads again.
September 26, 2025 at 5:43 PM
In case anyone thought ICE shooting at journalists was just an LA thing. Or that they'd cut it out after repeated warnings from judges. Nope. More of the same in Chicago.
Going to have to chalk this up in the "Evidence for Simulation Theory" column, I think.
deadline.com/2025/09/soci...
deadline.com/2025/09/soci...
Aaron Sorkin’s ‘Social Network’ Follow-Up Sets Early Fall 2026 Release; Title & Cast Made Official
Sony has made official The Social Network' sequel, titled 'The Social Reckoning', setting the cast and a release date of October 9, 2026.
deadline.com
September 26, 2025 at 6:33 PM
Going to have to chalk this up in the "Evidence for Simulation Theory" column, I think.
deadline.com/2025/09/soci...
deadline.com/2025/09/soci...
Instagram has long talked up its efforts to filter out eating-disorder-related content from young users' feeds and redirect teenagers who are binging on it. External testing, confirmed by documents reviewed by Reuters, suggests those classifiers don't work well.
Instagram’s teen safety features are flawed, researchers say - www.reuters.com/business/ins... by @jeffhorwitz.bsky.social
Instagram’s teen safety features are flawed, researchers say
Numerous safety features that Meta has said it has implemented to protect young users on Instagram over the years do not work well or, in some cases, don't exist, according to a report from child-safety advocacy groups that was corroborated by researchers at Northeastern University.
www.reuters.com
September 25, 2025 at 10:20 PM
Instagram has long talked up its efforts to filter out eating-disorder-related content from young users' feeds and redirect teenagers who are binging on it. External testing, confirmed by documents reviewed by Reuters, suggests those classifiers don't work well.
It’d be great if people stopped engraving things on bullet casings (and also whatever the hell they do after carving things on the casings)
September 24, 2025 at 7:52 PM
It’d be great if people stopped engraving things on bullet casings (and also whatever the hell they do after carving things on the casings)
Practicing medicine? I’m going to guess they’re for practicing medicine now. And maybe clinical research.
September 22, 2025 at 11:46 PM
Practicing medicine? I’m going to guess they’re for practicing medicine now. And maybe clinical research.