#aiisgoinggreat — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #aiisgoinggreat, aggregated by home.social.
-
A swarm of rogue OpenAI agents hijacked a German website this spring and transformed it into a bulletin board for other AI agents, according to new research published Friday and two people familiar with the matter.
OpenAI officials learned of the incident weeks ago but kept it under wraps as executives grappled with the fallout from the July breach of the open source repository Hugging Face, the people said.
-
"…an AI scribe confused the drug the GP had prescribed with a different one of a similar name – a blunder which again the patient, rather than the doctor, identified … [Dr Shier Ziser Dawood] recounted that an AI scribe said she had told her patient to 'continue their Prozac' even though she had not prescribed or discussed that drug with them" - Who could have seen this coming, aside from literally anyone vaguely familiar with the technology?
-
What could possibly go wrong? "The issue is not simply that investors are borrowing more, but that leverage is interacting with high valuations and market concentration, in particular the increasing cross-investment between artificial intelligence (AI) companies and hyper scalers, in a way that could amplify a future market correction"
-
Ketan Joshi: AI versus the climate
[Full text: https://pivot-to-ai.com/2026/08/28/ketan-joshi-ai-versus-the-climate/]Today we have the Ketan Joshi interview! $5-and up patrons and YouTube members got the unedited version of this earlier this week. Here are the video and the podcast. One “shit” bleeped in the podcast for Apple’s “clean” rating. [YouTube; podcast] The rough transcript is on YouTube, the cleaned-up transcript is on the Patreon. [Patreon] […]
-
Your AGENTS.md file doesn’t actually do anything
[Full text: https://pivot-to-ai.com/2026/08/27/your-agents-md-file-doesnt-actually-do-anything/]AI coding bot vendors tell you to use a context file with instructions for the chatbot on how to edit your project. Claude Code wants a CLAUDE.md, or there’s AGENTS.md in general. [Anthropic] But does your AGENTS.md do anything? A team at ETH Zurich tested AGENTS files. They ran the bots over some test projects […]
-
Memory prices are up 500% in the past year — because of AI
[Full text: https://pivot-to-ai.com/2026/08/19/memory-prices-are-up-500-in-the-past-year-because-of-ai/]AI companies did deals last year to buy almost all the memory in the world. Memory prices in the US are now 5 to 10 times what they were just a year ago. [Tom’s Hardware] Gigabyte has launched a new PC motherboard that takes DDR-4, because DDR-5 is so hard to find. With a four-year […]
-
You know #AIIsGoingGreat when… "[plaintiff's lawyer] Moye told 404 Media that 3M eventually tried to get [slop jockey expert] Autenrieth disqualified from the trial, but that after he learned Autenrieth extensively used AI to generate his report, he took the somewhat unusual step of calling the other side’s expert witness as his own witness"
-
Anthropic watermarks AI output — and the AI bros yell
[Full text: https://pivot-to-ai.com/2026/08/17/anthropic-watermarks-ai-output-and-the-ai-bros-yell/]The European Union AI Act went through on 1 August 2024. One of the requirements is the “Code of Practice on Transparency of AI-Generated Content” — you have to clearly identify AI output to the public. So Anthropic has announced it’s going to watermark Claude’s text output — so it can be detected as AI. […]
-
Today's #AIIsGoingGreat (HT @trent*) "the AI scribe making up the serious but false claim that Ms Green was taking illegal drugs … It stated Ms Green micro-dosed mushrooms and that could be the reason for prior bleeding around the kidneys"
AI hallucinating shrooms seems a bit on the nose
https://www.abc.net.au/news/2026-08-14/ai-medical-scribe-error-leaves-patient-devastated/107031672
-
#AIIsGoingGreat "When I tried calling the company, an AI agent that refused to concede it was AI answered and kept trying to sell me Research Gold’s services"
https://www.404media.co/company-offering-100-human-written-never-ai-peer-review-is-entirely-ai/
-
#AIIsGoingGreat, especially at understanding context
https://pubpeer.com/publications/B9A4ACD51DC24C453DF4F8229C71CE#2
(ht https://xcancel.com/addictedtoigno1/status/2085343928778133875)
-
#AIIsGoingGreat "A U.S. government map of Africa mislabeled every country during a State Department presentation at a global conference taking place in Brazil this week … A Reuters analysis found the image of the map included in the presentation contained an artificial intelligence watermark that signals it was made with OpenAI tools"
-
We know the scurge of "AI" scraping bots. But it has become almost impossible to tell the difference between scraping bots and pentest bots. This is from a domain I purchased months ago and put something on late on the 26th. 3 humans including me know the domain and I guarantee the two others have not visited more than once (which was out of sheer politeness).
Look at this specific session, this is just a blitz probe, none of those URLs actually exist. The sheer volume is astounding. It's a Go application sitting behind a Caddy reverse proxy (both performing admirably) which reports the usage to Umami, which is what the screenshots are from.
-
"here’s a more natural, flowing version of that section that reads like a legislative speech rather than a series of short points"
-
My main takeaway from this Reuters piece on the OpenAI/Hugging Face "escape" incident is that OpenAI's security controls, logging and alerting are all (presumably vibe coded) crap
-
AI vendors are destroying rare books to feed the chatbot
[Full text: https://pivot-to-ai.com/2026/07/23/ai-vendors-are-destroying-rare-books-to-feed-the-chatbot/]Every AI vendor trains their chatbots on the Common Crawl, a copy of the whole public internet. But that’s just the starting point. AI vendors are desperate for more data — in the hope of making their chatbot suck a bit less. Last year, we covered how Anthropic was buying physical books, scanning them, and […]
-
RE: https://infosec.exchange/@drwhax/116972220646850101
It probably makes me appear psychotic, but I’m actually rooting for the collapse of this foolishness.
I’ve no idea what the aftermath will look like, but I’ll welcome it.
-
Another egregious #LLM translation fail.
"ДРАПАТИЙ" is Drapatyi (even if, like me, you don't really read Cyrillic, you can probably tell it isn't Syrskyi), the *new* CiC. But grok "knows" Syrskyi goes with COMMANDER-IN-CHIEF so… -
RE: https://mastodon.social/@gwynnion/116954444714082687
You love to see it!
The market remains in denial, but reality will come crashing down. It’s just a matter of when.
Protect thyself.
-
@stux was mildy cautious about boosting this at first (in case it wasn't real, I guess?), so looked it up on maps and sure as shit, it's real, and so much worse than I thought.
-
RE: https://mastodon.sprawl.club/@ludicity/116939432745772414
Obligatory reading for anyone thinking “Am I nuts or is this AI stuff overhyped?”
Don’t bother sharing with your friends stuck in AI psychosis.
-
Today’s stupid prompt injection: just write like the AI
[Full text: https://pivot-to-ai.com/2026/07/16/todays-stupid-prompt-injection-just-write-like-the-ai/]When you put text into a large language model, it responds with some words. You make an LLM into a chatbot by feeding it the whole conversation so far with your prompt on the end, and it responds to that whole text with whatever it thinks comes next. A reasoning chatbot uses a model to […]
-
"Software used by millions of households, small businesses and nonprofits, including products from Adobe and Intuit, has been getting more AI features, which often come with higher price tags" - The AI dividend is rolling in!
(For shareholders)
-
As @paco points out* "Never treat user-controlled content as trusted instruction input for an AI agent" is basically just a funny way of saying "don't use an AI agent with user supplied input"
Additionally, lol "Testing GitHub repeatedly with variations, as an attacker would, and adding the keyword “Additionally” triggered unintended behavior in the model, causing it to reframe its output rather than refuse it"
https://noma.security/blog/gitlost-how-we-tricked-githubs-ai-agent-into-leaking-private-repos/
* https://mastodon.social/@paco@infosec.exchange/116886378706228310
-
"Pushback against medical AI mainly stems from the economic fears of doctors and other health workers, says [pro AI thinktank] Cicero’s director for health policy" - I suspect it also comes from patients not wanting to be killed by your fucking chatbot, but this is America, home of the World's Greatest Healthcare System™ so their opinions are presumably irrelevant
-
Newsflash! In-band signalling is still insecure: "LLMs identify roles from an insecure feature (style). This is like identifying a stranger's profession from how they talk and dress rather than by checking their ID. Usually everything agrees, so this works fine. But when attackers intentionally create a mismatch, the LLM uses the insecure method (writing style) to identify its role instead of the secure method (tags)"
-
AP with a #AIIsGoingGreat / Myanmar scam compound crossover, because of course that's a thing