You will also notice that 21/49 notes explicitly refer to a fact-checker. This a reminder that you can't just starve the ecosystem of professional journalistic endeavors and hope that the crowd will do its job instead.
Here are 49 notes flagging that a tweet was posting AI content about the Iran war without disclosing.
As you will see, most did not reach "helpful" status and will therefore probably not be used to demonetize creators spreading disinfo.
The policy may have decent precision, but not good recall.
In today's free @indicator.media I do my best to set aside the hypocrisy of X launching a fact-checking policy and the State Department applauding it -- and assess the merits of the policy itself.
This feels a little oversold based on the recall in the underlying paper, but only a little.
Mass deanonymization is a legitimate concern especially because of the cost, speed and scale at which LLM agents can operate.
Official statements placing U.S. forces in the area – along w/ analysis of social posts, videos, sat. images – suggest U.S. forces bombed a school in Minab, Iran during strikes targeting an IRGC base. Over 150 were killed, Iranian officials say. www.nytimes.com/2026/03/05/w...
I remain of the opinion that marketplace is the only good Facebook product
right. and there's also metadata, which as we've shown on Indicator does not reliably carry over
I also can't seem to find any more details about the policy on the policy pages... Typically these things would be associated to a blog post and a policy page update
help.x.com/en/using-x/c...
help.x.com/en/rules-and...
help.x.com/en/rules-and...
I hope this puts to rest any absurd assumption that the politically-motivated carping about "fact-checking/Trust & Safety is censorship" was in good faith
Quick, someone call Jim Jordan, Americans are being censored by the government
Alexios Mantzarlis, director of the Security, Trust, and Safety Initiative, is drawing international attention with his work.
His research, highlighted by @thestar.com, uncovered a massive AI podcast network using original work without credit. @mantzarlis.com
Learn more:
https://bit.ly/4ub92g0
what stage is this in @masnick.com's content moderation learning curve?
and so here's the helpful % going down. disproportionately for AI contributors tho
yes, take this with a pinch of salt as I only pulled it very fast and haven't QA'ed but looks like notes being written has gone up, helpful notes is stable-ish to declining
not a bad hypothesis! but overall numbers are going down too so i'm thinking maybe it's a rating side question of fewer people agreeing on what even is helpful
Interesting:
Contrary to my prediction, the share of AI notes on X appears to be dipping down. The trend has been notable and substantial over the past week.
indicator.media/ai-community...
All I can think about is Jeff Bezos saying that he slashed jobs at the Washington Post — including those of foreign correspondents in the Middle East — based on "the data," and a few weeks later a whole war breaks out in the Middle East that these laid off folks would have covered as true experts.
Check out the latest episode of @indicator.media Show & Tell!
Maldita.es has been doing some incredible work investigating scams, disinformation, and worse on the biggest tech platforms.
So Craig and I got @cjimenezcruz.bsky.social to walk us through three of them, spreadsheets in hand.
For those thinking of using Telegram for conflict reporting, image gathering, and source searching, here’s a guide I put together for @indicator.media:
indicator.media/p/the-indica...
whoever designed google cloud's dashboard is chaotic evil
Toronto's @thestar.com sent a cease and desist letter to the creator of the podcast network that @indicator.media wrote about on Wednesday. He has promised to reverse course.
www.thestar.com/opinion/publ...
Also excellent independent outlets:
indicator.media
www.wheresyoured.at
www.courtwatch.news
realhackhistory.org
www.londoncentric.media
privacyguides.org
databreaches.net
Not sure who needs to hear this but you can't use ChatGPT for realistic descriptions or images of Neanderthals.
(there's a serious takeaway: unequal access to an evolving knowledge basis can steer AI off course.)
GenAI is why Kant developed the Categorical Imperative.
Ask: What if everyone did this? Could the world still function? Would I accept that?
And here is the best example in living color 11,000 episiodes per day of AI slop.
indicator.media/p/this-ai-ge...
"For scammin' knowledge" is better than any slogan @craigsilverman.bsky.social could ever come up with on our own for @indicator.media.
I love it. (And thank you for your support)
This AI-generated podcast network publishes 11,000 episodes a day. It also rips off local media outlets.
indicator.media/p/this-ai-ge...
I went to see Operation Mincemeat last night and it was glorious. Highly recommend.
(It was also the new cast's first performance *and* Kids' Night on Broadway so the vibes were A++)
That is my one silver lining. I don't think many people would. But it does crowd out the good stuff
thank you! this is within the app?
