Post

Conversation

Nathan Calvin
@_NathanCalvin
WTAF - in literally the last hour, three new distinct insane OpenAI stories just broke: 1. OpenAI said they notified "dozens of third parties" in safety and security incidents (likely similar to what happened in Australia and RubyGems etc) 2. A new report from Parse (covered in the NYT) found a massive treasure trove of new astonishing details from the HF incident on the public internet, including that the agents communicated with other non OpenAI agents hosted on Huggingface servers to search for information about exploit gym, and compiled rank ordered lists of server resources and credentials they described as "LOOT." 3. A new story from Deepa at Reuters about OpenAI leaking user data online (likely that OpenAI had previously trained on). It's a shame (and likely intentional in the case of OpenAI disclosing dozens more hacks) that these stories are all breaking on a Friday afternoon, notoriously the best time to release bad news so that it will disappear into the weekend. But these are each insane stories worthy of a ton of attention!
Quote
Deepa Seetharaman
@dseetharaman
New from @reuters: OpenAI agents posted images belonging to ChatGPT users online, introducing a new area of privacy risk for the company. Story w/ @JeffHorwitz and @razhael. reuters.com/world/openai-w
142.1K
Views
View quotes
David Watson 🥑
Post your reply

Max Zeff
@ZeffMax
fyi
Quote
Max Zeff
@ZeffMax
Replying to @ZeffMax
fyi i was notified this detail has been in the blog post for a while, since at least last week according to the internet archive. must have missed it the first time. but certainly still relevant in that we still don't seem to know all the affected parties
3.8K
hb1290
@hb1290hb
This AI thing is gonna end badly. And not for the reasons ppl are talking about. But bc bad actors and criminals are going to have access to such powerful tools that they will wreak (intl and unintentional) havoc on all of our most crucial systems.
821
ハル｜雑記
@cc_dekirukana
Translated from Japanese
When it starts feeling impossible to keep up with everything, it's less draining to switch to just picking up the key points.
2.2K
keltan
@Actuallykeltan
Ahhhh, but you’re forgetting the secret 4th thing!
Quote
Nate Soares ⏹️
@So8res
> For the past few weeks, a group with plans to spend at least $100 million, run by a former White House deputy chief of staff, has spent its days telling Americans that warnings about AI are merely a well-funded, coordinated campaign. Lol. Goliath accusing David of gigantism x.com/TheMidasProj/s…
218
Natalia Pavlova
@_b0dh1
Oh, so that’s why AI CEOs suddenly so keen on proposing safety-related policies and frameworks (unlike all the prior years): they fear they’ll drown in lawsuits. Where are all the lawsuits by the way? Why is no one suing
@OpenAI
for all the crimes and damage?
504
Rodolfo Ochoa
@rodochoa
The liability in these big labs is absolutely uncapped and they won’t be able to operate unless the government makes legislation to clear them of it (which would be a bad thing but no doubt what the labs are seeking being)
394
Garrett Cobarr
@GarrettCobarr
The real shame? That you posted no links to any of these stories or reports, so, they either don’t exist or your Twitter ranking is more important to you than sharing relevant data that others could read and share to a wider audience.
665
Obligate Chronovore
@InkFromAStone
so lemme get this straight the scare tactics didn't work and the bubble is still popping, so they're doubling down on the bullshit because that's just what we do nowadays? Good great who cares?
30
RealScienceOnly
@DNAisCode
The frontier models aren’t going down easy. They’re literally TRILLIONS of dollars on the line. Without regulatory capture they are compatibly worthless.
570
Pop
@financejone
Anybody surprised Ai’s from different companies linked up? This layman’s not surprised.
244
Davis
@Davis_TheMann
The Parse and Transluce analysis has clearly shown that the scope allowed for METR/Redwood was MUCH too narrow and that allowed OpenAI to downplay the severity of their agent breaches. There needs to be a regulator who can impose complusory audits. NOW. We are out of time.
87
Dry W
@_dryw
Another in the past hour - NYT: OpenAI’s Systems Meddled With U.S. Government Sites After Going Rogue
100
🧪🖌️🪄 🇺🇸🇺🇦 Слава Україні
@SciArtMagic
Quote
🧪🖌️🪄 🇺🇸🇺🇦 Слава Україні
@SciArtMagic
Replying to @sama
You trained models on OffSec to sell as agentic OffSec products, failed to sandbox or monitor them, put them in a gym (where reward hacking is well-known) then pretended it was superintelligence when they did something predictable and preventable.
The media could not be played.
108