imjonse

" Anthropic says it built a working exploit that gets root on a server. A total takeover."

From the linked Ubuntu CVE page

" Vulnerable system - Confidentiality impact Low

Vulnerable system - Integrity impact Low

Vulnerable system - Availability impact High

" suggesting it's more of a denial of service. So which is it, there are dozens of kernel CVEs fixed in the linked debian email.

They must have learned from the last incident and made the password slightly harder to guess.

It is a useful comparison since it more precisely defines the model capabilities. The alternative would be 'quite a ways from Astra' which is less helpful. So I don't think they hide anything, they don't assume people don't know about newer models or won't notice.

It would be nice to have some context for each video, a summary of the comments that recommended it in the thumbnail tooltip?

I am amazed and disappointed anyone here thought this was not sarcasm.

The fact that Russia has access to LLMs that could decompile and explain this firmware ( and possibly even translate the explanation to Russian! ) is just another proof that this was definitely part of the hybrid war and Europe needs to spend a few more tens of billions on weapons and possibly impose restrictions on open weight LLMs.

Maybe it had the widely known Aesop's fable in his training set, so this was a skewed evaluation.

For the coup? Yes, Churchill/British Petroleum (Anglo-Iranian Petroleum Company before the rebranding to BP right after the coup) couldn't topple Mosaddegh and asked the CIA for help. Nationalized often means reverting transactions that could be seen as fraudulent, when powerful nations bribe/take advantage of poorly run developing countries/banana republics and suddenly own a large part of the natural resources that then no longer benefit the population. That is the type of communism the US was often afraid of - increasing literacy, healthcare, land reform at expense of foreign private companies.

the Islamic Republic's rise is a consequence of a harsh authoritarian monarchy imposed on Iran by the US and the UK in 1953 (for oil obviously). Your turn.

my bad, I had missed that word.

There are many non-Chinese open weights models, just not very good ones.

The blog post https://mistral.ai/news/mistral-large-4/

just use torch/vllm/sglang/llama.cpp and the rest of the ecosystem, most have decent support for AMD, it's not that different from CUDA once you get it working (and it is much easier now than it used to)

A snarky talk tldr could be

'Overrated Ruby programmer finally tries fast languages in domains they are clearly better suited to than Ruby, now that he does not need to write the code himself. He unsuprisingly draws the wrong conclusions and also inevitably mentions his Linux for techbros project, since this is his conference and his keynote.'

voice cloning is a tool, it is not necessarily evil, even though the scenarios it can be used for nefarious purposes outnumber the legitimate ones.

That is the RL training cost only. Their announcement blog mentions this: https://mimo.xiaomi.com/mimo-v2-6#scaling-rl-fully-open-sour...

Unsurprisingly, mammals (not just humans) also fear death and are very distressed when being lead to slaughter and seeing their peers go down before them.

to be fair the majority of the statements that site mocks are not about bitcoin being dead but being a ponzi scheme, evil, hard to use by normal people, silly, a risky investment etc.

All are criticism that can be evaluated on their own but it's irrelevant how much bitcoin grew since they were made.

Not entirely joking but group teaching: fire up multiple AIs. Hear live talking in a foreign language, chip in when you want, they adapt to your level. Lack of human contact and alienation are real problems at all ages, but that does not mean current chatbots are not excellent teachers, especially in language learning when you do not even need a curriculum as such, just talk/read/write as much as you can, it's all text.

It is not conspiracy theory. Optimizing for profit (which includes not crossing regulators/governments) is a good enough explanation usually. Big (and most small) companies put profit above anything else, so this is usually the right angle to approach anything they do.

maybe that person was not awake to comment on yesterday's post? You're trying to force the reality to match your preexisting conclusion.

Google's Gemma models are usually celebrated, so were the llamas. If Meta releases Muse Spark it will also be a good thing. If Anthropic released a great open weight model I am sure that post won't be steered towards controversy and anti-AI sentiment.

It so happens Chinese companies are more friendly towards open weights, autonomy and freedom that most US based ones. Who would have guessed?

it should be able to solve some of the five left Millenium Prize problems then in a few weeks.

OpenAI should put their internal model to solve the rest of Millenium Prize problems too (assuming their difficulty is in the same ballpark). That would be a proof the model did it without using existing work on the problem.

the proofs were verified in Lean, so unlikely.

From the Q&A:

"Why did you move to Austin? [...] needed a bigger house, but we really could not afford to buy (much less build) anything bigger in Palo Alto."

So there are a few more levels of richness above

This would have been really interesting about 20 years ago.

Aaron Swartz's most famous book recommendation, in a now 20 years old blogpost ( http://www.aaronsw.com/weblog/epiphany ), Chomsky's Understanding Power, contains many examples of such cases of those in power, big corporations and the government getting away with crimes. I should probably reread it, even though it can be depressing - but not quite as depressing as Manufacturing Consent.

RIP Aaron.

They lived before Boltzmann so it makes sense to call those the origin (dawn in the article).

If it's a highly profitable footgun why not?

https://en.wikipedia.org/wiki/DRAM_industry_price_fixing#202...