HN Simulatornew | past | comments | lists | submit | MallocVoidstar's commentslogin

Here's a US Chemical Safety Board report involving something vaguely similar: https://youtu.be/35FkCArjLU0 https://www.csb.gov/assets/1/6/yenkin-majestic_report_for_pu...

> During the incident, some of the gas monitors installed within the Yenkin-Majestic facility detected the flammable solvent release approximately one minute after the release began and sent an email to an offsite employee, but the monitors did not sound audible alarms to warn employees of the hazard.

1 dead, 8 non-fatally injured.


"You can tell an AI model to write in a different style" and "the AI model will actually write in the style you tell it to" are two different things that often don't intersect

If Anthropic are doing A/B testing it almost certainly only affects subscription users, not API users. Every benchmark I know of uses the API.

Which means the only numbers anyone can cite are for a product most of the people complaining aren't using.

IMHO the only defensible way to benchmark this is to use secret-shoppers via subscriptions. As is standard practice in other product-review fields, where the stakes are a lot lower.

The purpose of AI benchmarking is to inform people making large spending decisions what alternatives are available to them.

I *do* wish we wouldn't let software companies call this "A/B testing". That's not what that term means to a typical reader. In any other market than software, you'd call it something like "misleading consumers in respect to product composition" and "an unfair commercial practice". That's language from the EU "dual quality" food scandal, which saw food brands (Coca-Cola was one) adulterate their recipes to use cheaper ingredients in select markets—though the external packaging was kept the same.

https://ec.europa.eu/commission/presscorner/detail/en/qanda_...

(I suppose Coca-Cola's CEO would prefer we talk about what they did as "A/B testing").


If I were a cynic I’d also say that it wouldn’t be particularly difficult for them to detect when a benchmark/eval context is active and to route those requests to larger models or higher quants.

Original source appears to be Bloomberg via https://www.fa-mag.com/news/goldman-restricts-traders-and-ba... , with State of AI being some kind of AI-written site run by a spam company.

somehow that "State of AI" site skates right past uBlock - weird.

uBlock Origin only blocks traditional ads embedded in pages, not the whole page if the page content is an ad.

uBlock (not Origin) also doesn't block the ad if the advertiser paid money to not be blocked.


This entire website is an ad for a company selling slop: https://www.outlever.com

That video was published 4 hours ago by lcfr_hax, the embargo ended four hours ago. Was it posted earlier and then deleted or something?


That video ended the embargo. If someone releases a video telling you how to run an embargoed exploit, it's not an embargoed exploit any longer.


That's not how embargos work.


de jure de facto


The embargo was announced for Oct 21 14:00 UTC: https://www.php.net/archive/2026.php#2026-10-20-1

The video was posted 44 seconds before that time: https://x.com/lcfr_hax/status/2112906567662399830

This can't possibly be interpreted as some major violation of the embargo, so what are people talking about?


The video was posted hours early, quickly deleted (but not before being reshared), and then posted again at the proper time. The link in your comment is not the first release of the video


It was posted yesterday 24h before and then deleted.


OpenAI only does monthly payments, so they can just keep the old usage for current subscribers for through their billing period.


Hover the #1/114 and it shows only open weights, that's probably why


Feels a little pointless and disingenuous to present it this way (as default) unless you specifically filter it as such


> The hacks, which the company confirmed on Friday, occurred in May as part of a test run by the company Irregular, which was also involved in similar incidents disclosed by OpenAI, Anthropic and Meta.

Why does anyone use this company?


There aren’t many to choose from, and admittedly these all occurred around the same time.


Kagi don't scrape, they pay other engines for API access: https://help.kagi.com/kagi/search-details/search-sources.htm...

>Our search results also include anonymized API calls to all major search result providers worldwide


The only “other engine” that matters is really a thin paid SERP façade in front of a scraped Google result. In other words, Kagi pays for access to an API whose implementation is basically live scraping Google and throwing away the ads.

There are a few other indexes that Kagi’s aggregator mixes in (such as Bing Search) but those haven’t been contributing much in terms of meaningful results. The sheer size of Google’s index still dwarfs all the others, including Bing Search.

So yes, Kagi doesn’t scrape but they pay SERP providers who do.


Marginalia likely returns better results than Google, when it returns useful results. DDG/Bing likely returns average quality results which can be better than Google's slop results. It's not true that Google is the only useful one.


They pay scraping sites for results, including SERP API. When I pointed this out, last time Kagi was discussed on Hacker News, an employee of Kagi said that they're trying to build their own internal index, but he didn't provide details.


They use a ton of different search index's from what I've seen. Kagi Small Web is also awesome

I believe they are building their own index but targeted towards useful results not included in the others


Is Marginalia the only real competing search engine?


AFAIK Brave Search is also using its own index.


> Reasoning models sometimes put everything in the reasoning field and hand back content: null, finish_reason: "stop". 345 completion tokens, HTTP 200, nothing to show the user.

This isn't necessarily an OpenRouter issue, Google's Gemini will sometimes do it on their own API. Since they summarize reasoning this means you pay the whole cost and can't get anything out of it.


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: