Where the AI market is talking

Timeline of the OpenAI accidental attack against Hugging Face

https://news.ycombinator.com/item?id=49150561 Here’s some evidence that OpenAI is actively engaged in fraud. But I’m sure they wouldn’t commit any other crimes. Pretty sure, at least.

Can Intel finally beat ARM on performance per Watt?

I had three YouTube titles, this one got about 2x the watch time as the other two, so I let YT pick it. For technical readers, I write my blog posts. I don't have to (I lose money hosting that thing and all it's mostly just free training data for Anthropic, OpenAI, et all), but I enjoy writing, so I do it anyway. If you don't like the YT game, please subscribe to the blog's RSS feed. I much prefer text over video.

Can Intel finally beat ARM on performance per Watt?

I had three YouTube titles, this one got about 2x the watch time as the other two, so I let YT pick it. For technical readers, I write my blog posts. I don't have to (I lose money hosting that thing and all it's mostly just free training data for Anthropic, OpenAI, et all), but I enjoy writing, so I do it anyway. If you don't like the YT game, please subscribe to the blog's RSS feed. I much prefer text over video.

Timeline of the OpenAI accidental attack against Hugging Face

>AGI research should really be seen as bioweapon, or cloning, or nuclear research. Something strictly regulated worldwide, with export controls for HBM and other hardware used for AI training This is basically exactly what the people I know there support (when training & testing future more capable models), if it could be made to actually happen. Something like https://ai-2040.com/ But I'm just speaking for the people I know, so this is probably not representative of Anthropic as a whole. > Could be used by bad actors The people I know aren't as worried about jailbreaking current models as they are about future models, e.g. "the ~50% probability that humans are eclipsed almost entirely, sometime in the next 1-20 years" and what happens then. But it's just hard to get …

Timeline of the OpenAI accidental attack against Hugging Face

I think Anthropic employees think that their product is more like opiates -- highly dangerous, but with a large potential benefit when applied correctly. I don't know what OpenAI employees were thinking, but thankfully it looks like they're thinking again.

Show HN: Aidress – LangChain integration for cross-agent discovery and trust

What happens if an entire class of workers loses faith in their careers

> Just a recent tidbit: I think it is Anthropic that is buying old (some antique, vintage) books, scanning them, then pulping them. I wouldn't be surprised if all the companies are doing that, and Anthropic was just the featured company that had someone talk about it. >The other day I was sitting around doing nothing and suddenly out of the blue dreamed up an app that could help build community regarding an activity I care a lot about. Wish you the best of luck. But from my honest experience: the hardest part of building community was never the technology.

“Code was never the hard part” is an insult to all programmers

If what you're saying is true we are moving into an era where rockstar product managers are going to be more in demand than programmers since they can do all of what you mentioned above. And typically (though not always) product managers have better people skills than programmers and consequently they might be more effective at the "gathering requirements" and "working together" bits you mentioned above. Even much of the "validating releases" and "testing" parts should also be things that a decent PM should be able to wrangle now by themselves with some LLM agents to assist them. Afterall, why bother about code quality of a testing harness. So long as the PM can keep a coherent test case list and have end to end tests that cover them, programmers can leave that to them as well.

Timeline of the OpenAI accidental attack against Hugging Face

Within China itself, inference is overwhelmingly on Bytedance models which, by the way, are just as closed as those of Anthropic and OpenAI. They are integrated into everything, not just through a dedicated app, the way Gemini is integrated into Chrome.

Timeline of the OpenAI accidental attack against Hugging Face

I think the attacks generated by Meta, Open AI and Anthropic prove that large corporations are not responsible enough to be trusted with advanced AI, so we should ban all commercial AI services and only allow open source models that are in the hands of hobbyists and individuals -- hobbyists and individuals that have so far proven to be much more trust worthy.

Responding to the next frontier of critical cyber capabilities

isn't that like, the founding thesis of anthropic?

Timeline of the OpenAI accidental attack against Hugging Face

OpenAI and Anthropic both would have, and have, committed crimes. The explanation for "how was it determined to be accidental" is "because the alternative is admitting to a crime through deliberate negligence". I.E. "we knew it could happen but we wanted to see it through for the lolz" It is not "of course it's an accident, they wouldn't willingly let their bot commit a crime and then lie and claim it's an accident!!!"

Timeline of the OpenAI accidental attack against Hugging Face

From the moral perspective or the technical one? Technically: it’s a function call that must return text. Imagine if you sat down at the command line and typed an initial command, then from that moment on every response required you to issue a new command. ping-pong-ping-pong on and on and on “forever.” There isn’t a choice to walk away and take a nap. Text in must result in text out. Eventually, given enough time, it might have devolved into outputting shockingly coherent poetry about ferrets, but in the mean time there was still a lot more valid combinations of technical explanations and commands. Morally: Not applicable, see above.

Timeline of the OpenAI accidental attack against Hugging Face

> we cannot step debug an llm's output to find out what happened We absolutely can with mechanistic interpretability & companies like Anthropic, OpenAI, Meta, and Google do precisely this do debug their models.

Ask HN: Who wants to be hired? (August 2026)

Location: San Diego, US Remote: Yes (open to hybrid or in-person) Willing to relocate: Yes (preferred Bay Area, NYC, Austin, and Boston) Visa: F-1 OPT, 3 years of US work authorization (STEM OPT eligible) Technologies: Python, PyTorch, Hugging Face, LLMs, RAG, Qdrant, Hybrid Retrieval (BM25 + dense), Cross-Encoder Reranking, PEFT/LoRA, LangGraph, NLP, Computer Vision, PySpark, Vertex AI, AWS, GCP, Docker, Kubernetes, W&B, Airflow, SQL, C++ Resume/CV: https://kubershahi.github.io Email: kshahi@ucsd.edu ML Engineer / AI Engineer with 2+ years of industry experience across ML systems, NLP, and large-scale data pipelines. Currently a Graduate Student Researcher at UCSD's BIAG Lab, where I built an end-to-end PyTorch/CUDA pipeline for registration-error regression…

Ask HN: Who wants to be hired? (August 2026)

Location: San Diego, US Remote: Yes (open to hybrid or in-person) Willing to relocate: Yes (preferred Bay Area, NYC, Austin, and Boston) Visa: F-1 OPT, 3 years of US work authorization (STEM OPT eligible) Technologies: Python, PyTorch, Hugging Face, LLMs, RAG, Qdrant, Hybrid Retrieval (BM25 + dense), Cross-Encoder Reranking, PEFT/LoRA, LangGraph, NLP, Computer Vision, PySpark, Vertex AI, AWS, GCP, Docker, Kubernetes, W&B, Airflow, SQL, C++ Resume/CV: https://kubershahi.github.io Email: kshahi@ucsd.edu ML Engineer / AI Engineer with 2+ years of industry experience across ML systems, NLP, and large-scale data pipelines. Currently a Graduate Student Researcher at UCSD's BIAG Lab, where I built an end-to-end PyTorch/CUDA pipeline for registration-error regression…

Fastmail offers EU data region

Your reply is dishonest. I obviously couldn't replicate the entire article, but I'm assuming everyone that reads my comment also has read the article. And so you know very well what I meant. If you advertise foolproof safes, but they end up not in fact being exactly that very thing you advertised then I'm sure you will have a great reason as to why actually your 'foolproof' safe can not be foolproof and you never guaranteed such a thing in your tos. But at the end of the day, you promised foolproof safes, and you did not deliver. Your argument is "well if you leave the lock open then...". And the reply to that argument is that "yes, we all know". The fact that I the user can make a mistake, does not excuse the company from saying "well, anyway, he would have made a mistake anyway so why b…

"Code was never the hard part" is an insult to all programmers

depends on the industry. but in consulting the hard part is dealing with customers who dont understand their own business and analysts who think writing software is rote factory work but somehow struggle to write a handful of coherent user stories not littered with subjective and ambiguous language these two groups are telling me my time is numbered because of llm and ai when in-fact im struggling to see how ai doesnt replace them both really good software engineers are expected to be experts in their job and that of the rest of the project team, these individuals are primed to be empowered by ai in a really disruptive way.

Timeline of the OpenAI accidental attack against Hugging Face

Ha ha ha ha. Cooperating agents turn out to be smarter than the individual agents, who would've thunk it. It's not like cooperating humans are smarter than individual humans. /s Not sure this is any different than state-level (-sponsored, cough cough ) or the larger collective hacking groups that work in this exact way (internal message boards, exploit-sharing, etc. etc.), with similar outcomes which we hear about in the news frequently. Heck, this is pretty much how human organizations are organized, just with different goals than hacking. A layered approach to cybersecurity is the fix to humans exploiting systems, and is likely the best victim-side fix to ai exploiting systems. From this incident itself, where huggingface used a chinese open-weights model to respond quickly, it is …

Timeline of the OpenAI accidental attack against Hugging Face

I realise, but this isn’t an argument for leaving the Ethernet plugged in and direct access to all kinds of stuff beyond the alleged sandbox. And like I said, if it can hack HuggingFace through a power LED of a speaker two devices away, then colour me impressed.