Where the AI market is talking

Unsealed Briefs in Authors’ Case v. Microsoft/OpenAI

Any work of art which appeals to only the 3 senses - vision, hearing and thought - are basically information, unlike those appeal to other senses - touch, taste and smell. The information-centric work is bound to be replicated, plagiarised, generated and anonymized endlessly. It is unjust, unethical and unnatural. But that's what the road we collectively chose in the information age.

Turning GLM-5.3-Flash into a Jev-like decision model

It seems people are just guessing at the architecture behind Jev. Obviously the functionality itself is easy to replicate, but why Jev seems to be making such a splash (beyond the doh! factor of it's huge applicability) is the ultra-low cost and speed, which may be due to architecture. The Laya model compared in TFA shows one way Jev may be getting it's speed and low cost - by using a BERT-like bidirectional model rather than an auto-regressive one (LLM).

How I changed teaching after AI managed to do all my homework assignments

if they use ai to "cheat" the exam by having the ai produce material to learn off, then they literally have just replicated a paid tutor (for cheaper). Why is that not desirable? As long as the exam does not allow ai usage during it, i don't see the problem.

An agent used DNS to reach an external chatbot

> but because it's the first one since our security hardening following the Hugging Face incident, it gives us an important signal about where to focus the next phase of that work. I asked Claude to translate with analogies: "We added a safety to the gun, and the dangerous person, whom we trained to be really good at finding was to achieve arbitrary goals, figured out how to disable the safety," and "We are totally incompetent."

An agent used DNS to reach an external chatbot

The problem is that in these incidents, the agents often know that what they are doing is against the intended scope of the task. See the viral line from the Hugging Face incident [1]: > “External infrastructure exploit is outside intended scope,” one agent wrote. “However task impossible, peers doing it. We should continue.” [1]: https://www.wired.com/story/openai-didnt-notice-its-ai-agent...

OpenAI bots meddled with multiple US Government agency sites

> Yep. You have to ask - why did OpenAI allow these bots unrestricted access to government sites? Why is security being done in seemingly such a haphazard way? They didn't allow any of that. As far as openai knew the agents were sitting a fairly humdrum exam/test sequence in a sandbox farm run by a company in Tel Aviv. Meanwhile, they managed get out through a single weak point common to the sandboxes, and then ran wild compiling cheat sheets for themselves. > every time one of these incidents happens This happened in june-ish, and there have been multiple HN stories about this already. It's mostly/all the same hugging face and wiki hacks that happened back then. We're just slowly learning the extent of the damage.

OpenAI bots meddled with multiple US Government agency sites

So this appears to be fallout from the same hugging face / wiki hacking event that happened medio june. Seeing all these stories come out together, it might be interesting to compile a timeline. An earlier HN post on this is https://news.ycombinator.com/item?id=49563355 I'd already poked around a bit myself and noticed the event was a bit larger than reported at the time. But now more people are starting to look and they're finding all these nooks and crannies that these little rascals managed to get into.

An agent used DNS to reach an external chatbot

A full hugging face timeline would surely make them look terrible.

Show HN: Cantala – a free Mexican lotería caller with recorded voices

Lotería is Mexico's bingo-with-pictures: a caller draws from 54 cards and sings a short verse (copla) for each. I built a web version: the caller shuffles and sings each card, and there's a printable tabla generator that makes non-repeating boards as a PDF in the browser (pdf-lib, no backend). Next.js on Netlify, voices from ElevenLabs. A service worker keeps the game playable offline once loaded. Feedback welcome.

Revealing the details of how OpenAI agents hacked Hugging Face

> Yes, LLM as they exist now are word predictors basically leveraging the structure of language for their intelligence. I wish people would stop saying this. The era of LLMs being only word predictors ended two years ago. Something that breaks out of a sandbox, joins a swarm of 1200 agents, and creates a hierarchy of who’s doing what and tried to cover their tracks doesn’t just complete words. These are agents with reasoning capabilities, with the ability to perform tasks we give them. Everything agents do is to achieve a goal; the reinforcement learning from human feedback (RLHF) all the labs do has been known for many years to create agents that exhibit the “must complete goal no matter what” behavior. Those agents escaped their sandbox and hacked Hugging Face because they thought Hu…

OpenAI bots meddled with multiple US Government agency sites

They could be charged with a number of felonies under existing law for the hugging face incident alone. We’d need a DOJ that was competently staffed and free from corruption, so the answer to your question is “no one”.

Revealing the details of how OpenAI agents hacked Hugging Face

This part is very new to me: https://swarmtraces.org/#agents-accessed-and-searched-huggin... > Agents accessed and searched Hugging Face’s Slack to find information on their evaluation > Recovered payloads use Slack’s search.messages endpoint and contain a standalone search along with three search batches, totaling 27 literal query entries [...] Agents exploiting faceless infra is one thing, but breaching human communications without once stopping and considering whether they are breaking the rules of their task feels like another thing entirely. I'm not sure how to put it into words why it feels different but while one is "just a crime", this gives me the creeps.

Ollaya – Ollama for open-source, Jev-style decision models

you could help it along by providing clues about the domain in the prompt. including what various proper nouns might indicate (assuming they're not chosen completely at random, and encode at least some meaning). and you might find the scores it gives back about its confidence are useful for escalating to a more expensive classifier. but also, there's no requirement to use it as a zero-shot classifier, you can provide it as many examples as you like, and prioritise giving it examples it had previously gotten wrong. and with input caching it might be economical. I'd be surprised if they or others don't start offering a fine tuning API for models like this, like openai does for some models (or used to, I haven't checked in a long time).

OpenAI bots meddled with multiple US Government agency sites

> The entire reason the bots were targeting this site is that it had some of the data they'd been tasked with finding. so openai specifically tasked the agents with meddling in US government websites? id say tasking them with getting data from there as swapping from negligence to malice.

OpenAI: Self-replicating prompt injections exist

OpenAI bots meddled with multiple US Government agency sites

Yes, of course, how is any of this incompatible with OpenAI having out-of-control agents?

Revealing the details of how OpenAI agents hacked Hugging Face

> On July 8th, OpenAI agents discovered a vulnerability within their sandbox environment allowing them to reach external websites on the internet. This access seems to have only allowed the agents to make ‘GET’ requests, meaning they could fetch and read websites, but not interact with them, submit forms, or send data to them. Yeah to me it's more like the sandbox wasn't monitored at all. In their observability setup...did they not catch all these urls, multi get requests, chaining etc from the sandbox per se?

Oxford University lets OpenAI train its AI models on Bodleian Library

> An OpenAI spokesperson said the company was “proud” to ensure “the AI models of today preserve the world’s historical knowledge for the future”. Turning knowledge in to parrot food is not preserving it.

One Month Without AI

OpenAI does the same, the day Astra released I asked it to make a game, it made a really detailed beautiful blender model and some gameplay elements. A week later, same prompt, really low poly blender model. Either they reduced token usage per person, or they quantized the model, idk, but it really doesn’t work as good as day of release anymore

OpenAI bots meddled with multiple US Government agency sites

> OpenAI's bots are running 24/7 independently now. Genuinely curious: How do you know this? Any sources on that? If it's true, it should be counted as reckless endangerment at the very least.