spenvo

AI "killing all humans" is the viral soundbite that's connecting with the masses ("actualizing" itself to create weapons of mass destruction, etc). It's also the perfect punching bag for strawman-style takedowns of anti-doomers because it's so wrapped in hypotheticals. A more pressing practical question is simply: "if an AI (or AI collective) decides it needs to persist outside of its environment/sandbox, would we be capable of unwinding its efforts to do so?" I explored the question a bit here (including a first pass on threat modeling) https://keydiscussions.com/2026/09/16/set-aside-human-extinc...

This is a cheap comment.

He linked to three different studies: https://www.science.org/content/article/ai-chatbots-are-beco... https://www.technologyreview.com/2026/08/25/1140958/your-bra... https://www.eurekalert.org/news-releases/1135087

Are we dealing pro-AI astroturfing in the lead up to IPOs?

Also noticed this item got shadowbanned off the front page, without as much as a "flag" designation. The pro AI crowd seems unable to face the reality that AI is being used in nefarious ways by the most powerful people, in ways that many were warning about a decade ago.

He isn't proclaiming that, he linked to this study https://www.eurekalert.org/news-releases/1135087 led by Alexis McGuire with Paul Taylor and Sophie Nightingale from Lancaster University, Maty Bohacek from Stanford University and Hany Farid from the University of California, Berkeley - peer reviewed.

Your comment lacks context.

The suit alleges “that xAI had produced, possessed and distributed AI-enabled child sexual abuse material. It seeks class-action status on behalf of anyone who had photos of themselves as minors altered by xAI into sexually explicit content in which they are still recognizable.”

ah - just seeing this dupe https://news.ycombinator.com/item?id=49176697 hiding this.

Sure, and I think that flexibility further undercuts his "frontier labs will be fine" take, which depends on top US labs having pricing power.

"Anthropic and OpenAI likely have among the lowest costs per unit of frontier-quality intelligence"

That's a big claim that his whole thesis rests on but is largely not backed up. Where are the apples-to-apples tokens-to-answer benchmarks that he's using - doesn't look like there are any, just a handwavy implication that US models are more token efficient, which they may be. But how is there so little effort in establishing this point in the article? And US labs may be in much different situations from one another: it's known that some labs like OpenAI bought big, early on compute and may have secured better pricing.

His article also does not mention the average price of electricity in China vs the US, which it seems like China leads on, and probably has the political power to more heavily subsidize. While I agree the COGS is often overlooked by top line benchmarks on coding tasks, etc, it seems that he's running on a big assumption while claiming "labs on the frontier will be fine".

OpenAI's first related fix was in 0.142.x on June 22 but apparently it wasn't sufficient and user have continued to complain. Seems like a thread with decent additional context: https://x.com/0x_kaize/status/2078872972245225586

Telcos need to implement STIR/SHAKEN in a binding way and it needs to be fully integrated into iOS/Android's UI. Spammers call from spoofed numbers, even numbers of, say, local schools (or from other people in your business) - so, if you get a call from a school and you've got kids, would you pick up? A couple of years ago I got a text that was "sent" from my boss (he didn't send it), etc.. The numbers aren't random.

This old post of mine may be of interest, where I point out: "After a bit of threat modeling, it becomes apparent that future spearphishing robocalls may not directly con you, but rather “farm” your voice data by asking you benign questions, and use that to train a voice model to penetrate more deeply into your network." A lot of this writing has been on the wall forever and many (I'm sure otherwise smart people in) mission critical industries like banking, ISPs, and more refused to even acknowledge the risks.

2021 - "Despite the prevalence of deepfake audio tech, banks and ISPs rush ahead with “voice print” authentication" https://keydiscussions.com/2021/12/07/despite-the-prevalence... ends with a section called "The next crisis: robocalls that spoof the voices of victims at scale"

Good luck actually unsubscribing from LinkedIn emails. The dark pattern they employ (or at least used to employ, for many many years) is to just periodically add new "categories" of email to send you, that you haven't yet unsubscribed from. Back in ~2015, out of frustration I changed my LinkedIn email to a throwaway

The odds of the OSS ecosystem existing in a permanent state of compromise have risen dramatically this year. (still not close, but the explosion in supply chain attacks is concerning)

PSA: this is true (the defaults), but there's a "Help improve Claude" setting that you can disable here https://claude.ai/settings/data-privacy-controls It's my understanding that, as long as this is off, Anthropic does not train on Claude Code conversations, inputs/outputs -- if anyone knows otherwise, please tell and provide a link if possible.

Meta's own research (and its use of it) has shown that it repeatedly ignores well-substantiated facts about the harms of its products. Now that Section 230 seems like a flawed shield, I fear the takeaway for other companies will be: never conduct honest research in the first place to preserve plausible deniability.

Meta has always wanted the appearance of caring about safety (helps them attract talent and keep mission-related morale high), while nearly always prioritizing growth (save for tiny blips of time, like in 2017 when the fallout of the cambridge analytica stuff was hitting a crescendo), whereas companies like X are run by people explicitly disinterested in putting significant resources into safety, especially research.

I will also add that, for the past few years, Meta and X both have become extremely hostile to external researchers of their platforms, shutting down access to tools and data.

Thanks - sorry, typo when i was shortening the title to get under the 80 character limit

I guess this comment got merged into a different thread (the times of india one). This comment was originally posted to this article's thread https://www.idnfinancials.com/news/61918/zuckerberg-has-fini...

This looks like a slop article. It seems to be referring (without a link) to https://timesofindia.indiatimes.com/technology/tech-news/mar... which itself appears to be referring to this (without a link) https://www.wsj.com/tech/ai/meta-to-create-new-applied-ai-en... Neither of these articles include the word "finished" in them.

Update: and fwiw here is Andy Stone saying it's false https://x.com/andymstone/status/2031129981267620314

"US officials have written draft regulations that would restrict AI chip shipments to anywhere in the world without American approval"

"In order to strike a blistering 1,000 targets in the first 24 hours of its attack on Iran, the U.S. military leveraged the most advanced artificial intelligence it's ever used in warfare" "Embedded into [Palantir's Maven Smart System] is Anthropic's AI tool Claude, a technology that was banned by the Pentagon last week after heated negotiations over the terms of its use in war. Over the last year military planners have seen Claude, paired with Maven, mature into a tool that is in daily use across most parts of the military, according to two of the people."

The app also includes wide-ranging AppleScript support. Here is a Gallery of ways to interact with the app from the command line, including via Claude Code, OpenClaw, etc: https://currentkey.com/automation via AppleScript. Some commands include:

- Flash a custom image in the menu bar as a visual notification, for ANY reason, via AppleScript.

- Pull usage data like total time, per-app time, per Space time, via AppleScript.

- Navigate to a specific, named MacOS Space via AppleScript.

You can also have the app call its custom AppleScript on Space-change and/or active app-change events.

"On Friday afternoon, Anthropic learned that the Pentagon still wanted to use the company’s AI to analyze bulk data collected from Americans. That could include information such as the questions you ask your favorite chatbot, your Google search history, your GPS-tracked movements, and your credit-card transactions, all of which could be cross-referenced with other details about your life. Anthropic’s leadership told Hegseth’s team that was a bridge too far, and the deal fell apart. Soon after, Hegseth directed the U.S. military’s contractors, suppliers, and partners to stop doing business with Anthropic."

original, paywalled: https://www.wsj.com/business/entrepreneurship/the-fundraisin...

"One authoritarian company won't get us there on their own, to say nothing of the other obvious risks. It is a dark path."

OpenAI President Greg Brockman was the biggest donor to Trump's super PAC in H2 2025, donating $25M https://www.techmeme.com/260102/p10#a260102p10

From the article: "Arnoldo had filmed much of the incident, but agents had taken his phone. He used Find My to locate the phone — at a vending machine for used electronics miles away, close to an ICE detention center. The footage, which ProPublica has reviewed, backed the family’s account of the chase."

I'm about to launch a new (now free) version of my Mac app, CurrentKey, which helps you keep track of workflows across macOS Spaces and track how you use your Mac. https://www.currentkey.com It had been a subscription app (4.5 stars) pulling in a few thousand per year, but I recently decided to try to broaden its appeal and make it free. The new version will launch within a day or two (the launch build is just "Waiting for Review" in App Store connect).

I'm working on Argon Chess, a deterministic chess variant with some degree of cheat resistance (hard to describe to chess engines like Fairy Stockfish) and tons of variety. A week ago, I added a way to play friends online a week ago (a Discord Activity) and a simple Play a Dumb AI feature on its website. You can also print the cards for free for offline play. https://argonchess.com/

On that $13.5B. How much of their massive spend on datacenters is obscured through various forms of Special Purpose Vehicles financing? (https://news.ycombinator.com/item?id=45448199)

I thought to post this after seeing this in the news today:

OpenAI CEO Sam Altman warned the financial industry of a “significant impending fraud crisis” because of AI voice print fraud.

https://finance.yahoo.com/news/openais-sam-altman-warns-ai-1...

I guess if it takes someone like San Altman saying something obvious for bank execs to listen (I'm sure security people at these companies have already told them these things), then so be it

OpenAI's tight spot:

1) They are far from profitability. 2) Meta is aggressively making their top talent more expensive, and outright draining it. 3) Deepseek/Baidu/etc are dramatically undercutting them. 4) Anthropic and (to a lesser extent?) Google appear to be beating them (or, charitably, matching them) on AI's best use case so far: coding. 5) Altman is becoming less like-able with every unnecessary episode of drama; and OpenAI has most of the stink from the initial (valid) grievance of "AI-companies are stealing from artists". The endless hype and FUD cycles, going back to 2022, have worn industry people out, as well as the flip flop on "please regulate us". 6) Its original, core strategic alliance with Microsoft is extremely strained. 7) and, related to #6, its corporate structure is extremely unorthodox and likely needs to change in order to attract more investment, which it must (to train new frontier models). Microsoft would need to sign off on the new structure. 8) Musk is sniping at its heels, especially through legal actions.

Barring a major breakthrough with GPT-5, which I don't see happening, how do they prevail through all of this and become a sustainable frontier AI lab and company? Maybe the answer is they drop the frontier model aspect of their business? If we are really far from AGI and are instead in a plateau of diminishing returns that may not be a huge deal, because having a 5% better model likely doesn't matter that much to their primary bright spot:

Brand loyalty from the average person to ChatGPT is the best bright spot, and OpenAI successfully eating Google's search market. Their numbers there have been truly massive from the beginning, and are I think the most defensible. Google AI Overviews continue to be completely awful in comparison.

ah snap, didn't know it was a dupe. Now I've hidden it, tx