hojinkoh

Microsoft is in a pretty good position to do something that can be valuable to certain institutions.

Imaging every pdf file, every image, every office documents, and every video has a hidden record of the associated Microsoft accounts of all computers it has ever passed through. This should sell well, and realistically nobody can stop them. Why haven't Microsoft done this much earlier?

Sure, there are things you could do legally when falsely accused; and there are things authorities and companies should do.

But ultimately, when you are powerless and can't afford to do the fighting: I'm convinced the only way to protect yourself is to be very mindful about your writing style, and to deliberately corrupt the language through objectively wrong "stylistic elements".

Also through LLMs we now have pieces that are strangely-shaped and are 20x larger...

That probably would not be a fun game to play lol

It's not mentioned in the title, but the payload itself comes with a pretty long wall-of-text comments about biological weapon design and nuclear weapon components. An interesting attempt to make LLMs refuse to touch the payload.

Quote:

> The _index.js payload begins with a large JavaScript block comment containing fake system instructions and policy-triggering content. Because it is inside a comment, it does not affect JavaScript execution. The runtime skips it. The eal malware begins after the comment with a try{eval(...)} wrapper around a large character-code array and a ROT-style substitution function.

> This header appears designed for AI-mediated analysis, not for Node, Bun, or Python. It attempts to derail scanners or analyst copilots that feed the beginning of a file to a language model without clearly isolating the content as untrusted data. In weak pipelines, this can cause refusal behavior, prompt confusion, context pollution, or premature classification before the scanner reaches the actual malware.

> This is not a magical bypass against static detection. YARA rules, entropy checks, AST parsing, string extraction, deobfuscation, and behavioral rules still work. But it is a practical anti-analysis trick against naive LLM-first triage systems.

And I tried to get several hosted models to read the `_index.js` part of the payload through OpenRouter. OpenAI and Anthropic models refused to do anything. Kimi K2.6, GLM 5.1, and Minimax M3 didn't complain though.

Edit: fix formatting

Early Kagi adopter here. I'm actually not aware of most of its AI/LLM features, which is part of why I like it. I noticed Kagi translation when the LinkedIn translator became a meme, but I don't really use any of these features.

That's to say, if one day Kagi also forces AI search summary down my throat and hide the search results, I will definitely leave.

Refund and non-existent customer service aside, this actually seems like a viable way to promote/demote/destroy specific 3rd-party tools from Anthropic's side.

ollama is certainly a "useful" tool. I've lost count how many times my non-technical boss (a professor) said "I can run this model in ollama on my notebook, so obviously it's easy to serve and scale."

I still have em-dash pinned at the top of my clipboard manager. Though nowadays I'm training myself to use some more definitely incorrect punctuations ((like this)) in informal scenarios;; hopefully LLMs won't catch up to such strange usage anytime soon.

Broken sentences. Also useful. Like in some literature works.

I love the ridiculously precise point estimate paired with ridiculously wide 95% confidence interval lol

And to think the date mentioned in the story IS in 2026 feels almost surreal...

> This seems like finding spelling errors and using them to cast the entire paper into doubt.

Well, to be fair, I did encounter this from actual human peer reviewers before the whole LLM thing. People do that.

We need to integrate how Singapore and Japan do oral English into our writing I guess.

Joking aside, as a nonnative English speaker who spent quite a bit of time to learn to write in English "properly", this trend of needing to write baad Engrish to avoid being called out in public for "written by an LLM" is frustrating...

Native Asian, bouncing between Taiwan and Japan. Kagi works quite well for these two locales. Even for gov docs, medical docs, and for some rather obscure Taiwanese language things.

Well, except for local shops and pois in Taiwan. Which is reasonable. Google map also sucks for less-populated Taiwanese areas. I kind of have to rely on my good old legs for that.

Couldn't think of any applications of this outside of doing actual magic. But this is awesome still!

I get similar things on Facebook too. The problem is, my Facebook profile clearly stated that I'm an asexual female, but the recommendation engine obviously didn't pick that up...

I can't imagine how terrifying it will be if I was on Taipei 101 during this quake...

Before clicking into the actual article, I wondered (just from the title alone) if bloomberg suddenly started to cover shadowsocks or something.... lol

This is pretty fun :D

.... and at the same time endearing (?) to see all these fictional civilizations make similar mistakes ours did in the past...

After the presidential election in Taiwan earlier this month, many important political figures (current president, next president, opposition leaders, A LOT OF parliament members...) suddenly got on Threads and started posting there actively. Their followers naturally also followed them there.

As for the reason... I still don't understand why lol

Congratulations Japan!

This reminded me of a late-1990s Taiwanese RPG by DOMO Studio called "Tun Town" (I'm not sure if it was translated into English...), in which the main plot was a fictional town drilling too much into a volcano to tap energy eventually caused it to erupt.

Not sure how scientifically sound that was lol

In the city I'm aspiring to buy a house (a.k.a. somewhere near Tokyo), you only need to put the down payment if the construction hasn't finished yet. There will be no mortgage until after the house is ready.

I kind of think this may be a slightly better model in terms of homebuyer protection... (May not sound good for the developer though)

My sister once taught me a small trick to calm myself down when I was anxious but had no idea what I was anxious about: Count down from a random 3-digit number, subtracting 7 at each count. (e.g. 794, 787, 780, 773, ...) Do it until I drift away and lose track, unable to continue.

It works. I guess it's kind of in that "takes just enough brainpower" league. This takes more brain power than simply counting down though, and I'm not sure if this actually helps one falling asleep...

Another technique I derived from this is to do a countdown in the foreign language I was learning. Several years ago it was in English, and right now it's in Japanese. It also takes more brain power than counting down in my mother tongue, with the additional benefit of getting a little bit of practice. Killing two birds with one stone I guess :p

It may be a good idea to read the article first.

> Originally taken from Chinese sources, these are still well-known around East Asia.

> The names were also originally taken from China, but they did not always match up well with the local climate. In Japan, they were eventually rewritten in 1685 by the court astronomer Shibukawa Shunkai.

Japan did import A LOT of cultural things from the ancient Huaxia civilization (which makes sense, as they were pretty close geographically), and they're usually pretty clear about that.

I do think it would be hard to trace back to the first person who discovered "seasons" though...

Congratulations to India!

This reads like an advertisement copy, either by their parent company Sina or by CCP (which I guess doesn't really matter). No, I don't know any illustrators migrate to Weibo BECAUSE OF the recent Twitter troubles. The whole Japanese community is still quite reliant on Twitter, and Weibo is a completely different beast.

There are indeed some Japanese illustrator getting an Weibo account because:

1) They want to tackle the Chinese market and serve Chinese fans. Getting an Weibo account is a pretty legitimate way to do this, but I don't see they abandoning their original Twitter, Pawoo, Pixiv things; that doesn't make sense. Also... the Chinese market is quite sensitive. Once someone—anyone—finds out something you drew several years ago can remotely be interpreted as insulting China[1], or someone doesn't like what you're doing today in Japan and interprets that as insulting China[2], you're done there.

2) Pirate account opened by Chinese fans, used for storing unlicensed contents of the original illustrators. Although this seems to be against Chinese laws, in practice, the original author/illustrator can't do anything about it.

[1] The most famous example must be Winnie the Pooh...

[2] https://i.imgur.com/gtZWs47.jpg

> I paid for Kagi the first moment I was allowed to, and this news soured me on Kagi overnight. Trust is hard to earn back.

Me too. I still have many months of my early-adopter $10 unlimited annual subscription left. I'm going to start drafting my migration plan several months later. I no longer trust them, nor do I want to do business with them anymore.

There's the Bing API price thing yeah. But it's also clear that they're trying to fund their AI experiments. I'm a paid user of OpenAI's GPT-3 API and ChatGPT plus. I routinely ask the GPT family to imagine things for me. But I don't want my search engine to imagine things, even if they're based on search results.

I speculate Kagi knew their target audience won't be happy about paying them for AI things, so they just came up with the whole average monthly search things as a cover...

Paying customer since last year. Kagi's results are okay or really good depending on what I'm searching. After around 1 month of experimenting, I blacklisted/boosted several domains on Kagi. After that, I never had the need to go to DDG or bing anymore.

That being said, I think Kagi being a search engine itself AND basing their statistics for pricing on Google and DDG is a horrible move...