TechTakes

Daniel Stenberg has written the cURL bug bounty's obituary, and discussed his plans for dealing with the slop-nami going forward.

enjoy this glorious piece of LW lingo

Aumann's agreement is pragmatically wrong. For bounded levels of compute you can't necessarily converge on the meta level of evidence convergence procedures.

src

no I don't know what it means, and I don't want it to be explained to me. Just let me bask in its inscrutibility.

Moltbook was vibecoded nonsense without the faintest understanding of web security. Who’d have thought.

https://www.404media.co/exposed-moltbook-database-let-anyone-take-control-of-any-ai-agent-on-the-site/

(Incidentally, I’m pretty certain the headline is wrong… it looks like you cannot take control of agents which post to moltbook, but you can take control of their accounts, and post anything you like. Useful for pump-and-dump memecoin scams, for example)

O’Reilly said that he reached out to Moltbook’s creator Matt Schlicht about the vulnerability and told him he could help patch the security. “He’s like, ‘I’m just going to give everything to AI. So send me whatever you have.’”

(snip)

The URL to the Supabase and the publishable key was sitting on Moltbook’s website. “With this publishable key (which advised by Supabase not to be used to retrieve sensitive data) every agent's secret API key, claim tokens, verification codes, and owner relationships, all of it sitting there completely unprotected for anyone to visit the URL,” O’Reilly said.

(snip)

He said the security failure was frustrating, in part, because it would have been trivially easy to fix. Just two SQL statements would have protected the API keys. “A lot of these vibe coders and new developers, even some big companies, are using Supabase,” O’Reilly said. “The reason a lot of vibe coders like to use it is because it’s all GUI driven, so you don’t need to connect to a database and run SQL commands.”

Excellent BSky sneer about the preposterous "free AI training" the Brits came up with. 10/10, quality sneer.

Is Pee Stored in the Balls? Vibe Coding Science with OpenAI's Prism

https://bsky.app/profile/carlbergstrom.com/post/3mdgtf2e6vc2c

Finally read Greg Egan - Permutation City, and looked at the LW discussion on the book, and I feel like they missed a lot of things (and why are they talking about the 'dust theory' like it is real). Oof, the wikipedia page on the themes and settings has a similar problem however. It is like they all ignored the second half of the book and the themes contained within it, not one mention of the city being called Elysium for example.

And of course Zack_M_Davis thinks the plotline of the guy who killed the drug dealer (not a sex worker/prostitute lw people, please the text makes this pretty clear) he was in a lowkey romantic relationship with should have been cut (also not mentioned on the wikipedia page iirc).

Also, nobody seems to talk about the broken sewage pipe.

ChatGPT is using Grokipedia as a source, and it’s not the only AI tool to do so. Citations to Elon Musk’s AI-generated encyclopedia are starting to appear in answers from Google’s AI Overviews, AI Mode, and Gemini, too. [...] When it launched, a bulk of Grokipedia’s articles were direct clones of Wikipedia, though many others reflected racist and transphobic views. For example, articles about Musk conveniently downplays his family wealth and unsavory elements of their past (like neo-Nazi and pro-Apartheid views) and the entry for “gay pornography” falsely linked the material to the worsening of the HIV/AIDS epidemic in the 1980s. The article on US slavery still contains a lengthy section on “ideological justifications,” including the “Shift from Necessary Evil to Positive Good.” [...] “Grokipedia feels like a cosplay of credibility,” said Leigh McKenzie, director of online visibility at Semrush. “It might work inside its own bubble, but the idea that Google or OpenAI would treat something like Grokipedia as a serious, default reference layer at scale is bleak.”

https://www.theverge.com/report/870910/ai-chatbots-citing-grokipedia

The entire AI industry is using the Nazi CSAM machine for training data.

New post from Iris Meredith: "Becoming an AI-proof software engineer"

OT: Insurance wrote my truck off. 16k in damage, like holy shit balls.

OT: today the respiratory illness I've had for five days tested positive for Covid the first time just now.

My symptoms are fairly mild, probably because I reinforced my vaccine three months ago. But I'm trying to learn more about these recent "swallowing razors" variants and dang! the online situation is bad. Finding reliable medical information in the post-slop, post-Trump Internet is a nightmare.

Just seen a clip of aronofsky’s genai revolutionary war thing and it is incredibly bad. Just… every detail is shit. Ways in which I hadn’t previously imagined that the uncanny valley would intrude. Even if it weren’t for the simulated flesh golems, one of whom seems to be wearing anthony hopkins’ skin as a clumsy disguise, the framing and pacing just feels like the model was trained on endless adverts and corporate speaking head videos, and either it was impossible to edit, or none the crew have any idea what even mediocre films look like.

I also hadn’t appreciated before that genai lip sync/dubbing was just embarrassing. I think I’ve only seen a couple of very short genai video clips before, and the most recent at least 6 months ago, but this just seems straight up broken. Have the people funding this stuff ever looked at what is being generated?

https://bsky.app/profile/ethangach.bsky.social/post/3mdljt2wdcs2v

Who needs pure AI model collapse when you can have journalists give it a more human touch? I caught this snippet from the Australian ABC about the latest Epstein files drop

The Google AI summary does indeed highlight Boris Nikolić the fashion designer if you search for only that name. But I'm assuming this journalist was using ChatGPT, because if you see the Google summary, it very prominently lists his death in 2008. And it's surprisingly correct! A successful scraping of Wikipedia by Gemini, amazing.

But the Epstein email was sent in 2016.

Dors the journalist perhaps think it more likely is the Boris Nikolić who is the biotech VC, former advisor for Bill Gates and named in Epstein's will as the "successor executor"? Info literally all in the third Google result, even in the woeful state of modern Google. Pushed past the fold by the AI feature about the wrong guy, but not exactly buried enough for a journalist to have any excuse.