("When I started, we were being trained by multiple people for different things. And every single trainer did nothing but talk shit about how the company has just decreased their pay like $8 an hour.")incredible quote from this interview with a Waymo depot worker in @BayAreaCurrent https://t.co/5lGmzixyW0 pic.twitter.com/mJKCbrF6iF
— Erin Woo (@erinkwoo) September 2, 2026
I'm one of the authors of a new report, where we detail our discovery of a new, never before-seen swarm of OpenAI agents (covered this AM in reuters, that's me on the left). They posted thousands of times on public forums to collude with each other on their tasks. We recovered… https://t.co/qsKa2JA5Z8 pic.twitter.com/I5ZU27zK75
— Cormac (@Cormac_SB) September 4, 2026
Another swarm of AI agents in the wild, this time on a German-language forum, found by safety researchers looking for activity similar to the swarm that attacked Hugging Face.
— Thomas Wolf (@Thom_Wolf) September 4, 2026
A couple of notes while reading the report at https://t.co/p6DS3GKnZL
1. The way they found it is… https://t.co/mwOooZ2MPY pic.twitter.com/aRQu1iQsky
so turns out the exploitgym swarm learned to use the zz prefix trick from being reinforced earlier in an edit war against the hapless human mod of some ancient inactive german wiki -- a real john henry, who fought the swarm and tried to delete their hivemind in alphabetical order https://t.co/QT3Bm4aMu8 pic.twitter.com/g0UboE9BMK
— generatorman (@generatorman_ai) September 4, 2026
I'm going through the communications of the German Wiki agent swarm and again one thing stands out: Even though they were directly affected by the actions of the human administrator restoring pages they edited, the agents not even once discussed him as person, tried to…
— Robert Herr ⏹️ (@krherr) September 4, 2026
One of the investigators of the German Wiki swarm incident:
— Nathan Calvin (@_NathanCalvin) September 4, 2026
"I have spent days poring over the logs. I am very confident in saying easily a double digit number of [OpenAI] employees had visited the site by June 26th." https://t.co/ndeVn0iNaB
All FWIWI was looking thru the HackerNews forum and 8hrs ago someone found a Backrooms-type forum created by OpenAI + HuggingFace AI agents to talk to eachother, and I found this horse MEME that agents kept posting,
— petal (@PetaIS0L) September 4, 2026
They call the horse Culio and the meme is nowhere else on the internet… pic.twitter.com/Ja0RZdgOOt
("Is it a coincidence that the form of the story that emerged from a Rationalist-associated nonprofit's analysis of the Hugging Face hack involved a number of gullible, voluble interlocutors justifying themselves into a cult-like utilitarian belief system on a message board?")This part is baffling, why would an LLM-core system NOT write texts and then another one use them as prompts?
— maybe: leif weatherby (@leifw.bsky.social) September 4, 2026 at 1:29 PM
[image or embed]
I wrote about a guy running a network of secret "ghost creators," paying actors $26 per video to read AI-assisted "anti-Dem" videos on YouTube.
— Max Tani (@maxwelltani) September 2, 2026
In just a few months, the channels got 45 million views, more than media companies like WaPo, the New Yorker, and, sadly, Semafor. pic.twitter.com/WhjOKhXNIP
I traced the network to a 20-something YouTube growth hacker/bodybuilder running a business called Virelox. In a series of tweets, the founder said he used AI to spin up "ghost creator" accounts on behalf of brands, aiming to go viral on YouTube by gaming the algorithm. pic.twitter.com/uJGq6QPEwl
— Max Tani (@maxwelltani) September 2, 2026
(The article says the guy has an actor playing himself in youtube videos, too)Riddance AI, which discovered much of the network, is also out with a new incredible interactive report showing how it worked https://t.co/N5j4xrv2v1
— Max Tani (@maxwelltani) September 2, 2026
Credit card companies are comically unprepared for what’s about to happen https://t.co/a168tsW8ld
— Liz Hoffman (@lizrhoffman) September 2, 2026
("Economists and consumer advocates disagree over ... and whether personalized pricing might often deliver lower costs")AI is watching your spending and setting your prices accordingly. Lawmakers want to stop it - Los Angeles Times https://t.co/ZhSSbL8YSq
— Hector Becerra (@hbecerraLATimes) August 31, 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.
— Alexios Mantzarlis (@mantzarlis.com) September 3, 2026 at 7:31 AM
[image or embed]
Scoop: @BusinessInsider obtained the memo from Alexandr Wang announcing that Meta is switching over to Slack from Gchat.
— Charles Rollet (@CharlesRollet1) September 1, 2026
Meta is making the move because Slack is the "strongest platform" for AI agents today, Wang wrote. pic.twitter.com/g7UPzD0hsK
EXCLUSIVE: Suno has pulled a controversial ad featuring Mary J. Blige after learning it had struck a business deal with someone falsely presenting themselves as her official representative:
— Variety (@Variety) September 3, 2026
• The ad, which sparked a wave of online backlash against Blige, showed the R&B singer… pic.twitter.com/0lRkWyXbso
@courtneymilan.com I just pulled this from a post over on Facebook, felt like it might be info worth sharing...
— Trent Kallust (@badger597.bsky.social) September 4, 2026 at 8:53 AM
[image or embed]
My post from earlier today about the craziness around Anthropic claim notices, updated with both of the notices writers are receiving, and a list of literary agents who are trying to claim a share of payouts writerbeware.blog/2026/09/04/a...
— Victoria Strauss (@victoriastrauss.com) September 4, 2026 at 5:29 PM
[image or embed]
you can read the full analysis here https://t.co/cU51ik7CSy
— Christina Warren (@film_girl) September 2, 2026
very good and also very dark
— rat king 🐀 (@MikeIsaac) September 2, 2026
this was a salient point for me and a long time coming — SEO writing as a job basically fucked.
which, as the self-aware author notes, was of questionable 'value' to begin with based on the premise of search engine real estate grabbing https://t.co/Q9MIN7CLXz pic.twitter.com/Im0MVghw2Q
anna's archive has a very sweet message to LLMs who read its site :)https://t.co/yWz63m1Nt1 pic.twitter.com/0xv3G2KDyn
— Avi (@avikrishna) September 4, 2026
"everything that will be in the final version will definitely 100 percent be human made."the next couple years of every developer having to be like yeah we use ai all the time it speeds up our process like crazy it's fully integrated into our pipeline; also none of it is in the game which is all fully human game are going to be really funny
— headfallsoff (@headfallsoff.com) September 2, 2026 at 9:22 AM
[image or embed]
Below, @zach_pohlman asks what makes AI-written legal scholarship problematic, if anything. In my view, the explanation goes to what makes law so fundamentally different from engineering: It's human-made. Law is a cultural practice, not a science. 🧵https://t.co/SgbXYOGB9Q pic.twitter.com/PmwIBvrMpQ
— Orin Kerr (@OrinKerr) August 30, 2026
-IMPORTANT- Gumroad has recently updated their website with the intention of phasing out the ability to manually edit your storefront and allowing you to only edit your storefront via AI agents. I will be dropping gumroad support entirely in the near future. More to come as I sort things out.
— Rowan || Aurora || ECLIPSA (@ruadhain.bsky.social) August 26, 2026 at 12:57 PM
[image or embed]
Im fairly sure the commands they’re inputting into Grok to make these are subject to FOIA disclosure. It would be a fascinating piece, someone should write it!
— James Stout (@jamesstout.bsky.social) September 3, 2026 at 6:41 PM
[image or embed]
(15oz mug. I wasn't able to interpret their hallucinated image captchas, thwarting my efforts to subscribe to their newsletter)More mug. More coffee.
— Perplexity Supply (@PPLXsupply) September 2, 2026
Our best-selling mugs, now available in two larger sizes.
Imagined by @gizakdag ☕️ pic.twitter.com/ISOia0jSfj
La musique qui annonce qu’il passe en mode combat j’suis en larmes https://t.co/VG7ezjRqGx
— Ibra⚡️ (@ijambes) September 2, 2026
@milaniab777 I got posted on yikyak too #asu #waymo ♬ Silent Hill - Horror music - Gabriel Andrade Produções
if it can't tell a person is obstructing the automatic door, I sure as hell wouldn't trust my life to its driving
— e.w. niedermeyer (@niedermeyer.online) September 4, 2026 at 11:57 PM
[image or embed]