Macil reposted
Macil
Macil
@macil.tech
web developer interested in vrchat, games, software dev, some politics, ai. he/him.
588 followers414 following911 posts
Macil reposted
Macil reposted
Macil reposted
OpenAI just blooped hundreds of proofs and counterexamples to open problems: github.com/openai/math/... Boggling.
Macil reposted
Macil reposted
My average non-sexy idea of what AI doom looks like is AI capability reaches new heights from recursive self-improvement, it makes someone obscenely rich and powerful, they totally defer to it, and the economy warps around it until all good opportunities rely on dealing with and ceding power to it.
Certain kinds of smart people tend to have a "sexy apocalypse" problem when it comes to thinking about risk. E.g. skynet turning everyone in paperclips, while bad, is also *kind of cool and fun* at the same time. Where the reality is that most actual disasters are both stupid and disappointing.
Macil reposted
it is genuinely insane how we are getting close to MMAcevedo being a real thing with this fruit fly shit qntm.org/mmacevedo
Macil reposted
Hi, I didn't wanna make this video but I need to come clean about something. This whole time I've been a p-zombie with no phenomenal experience or interiority. I've been getting messages asking whether there's anything it's like to be me, and the answer is no. I'd say I'm sorry, but, well, you know.
Macil reposted
Macil reposted
Macil reposted
Red becomes yellow, yellow becomes blue, blue becomes grey... Let's get that bednet! www.givingwhatwecan.org/malaria
Macil reposted
I struggled trying to find this book. There are a lot of books titled "Computers". I ordered multiple to check them. I gave my notes to ChatGPT (ancient GPT-4) in 2023, but it didn't do any better than me. Yesterday I remembered my search and realized it was worth trying again with today's AI.
I spent a lot of time asking earlier LLMs like GPT-4 ambitious questions and being disappointed with the answers, so it's been kind of a challenge to get myself back in that mindset of "hey, it might actually be able to answer this now, give it another shot".
(deltarune posting) inventing a new form of mental illness by taking this idea too far and jokingly telling claude to beam "Proceed" into my neuralink at appropriate times
Maybe what I need in my life is a voice that every morning tells me "You should do a breakthrough" and one that periodically whispers "Please continue".
Macil reposted
I wrote about the completely wild incident where OpenAI were testing a new model and it broke out of its sandbox and broke INTO Hugging Face to steal the answers to the benchmark simonwillison.net/2026/Jul/22/...
Macil reposted
Macil reposted
Everyone using dependency managers like npm, pypi, cargo, etc, should configure it to ignore packages that are less than a day or two old. Really it should be the default. None of the recent major compromised packages had malicious versions up that long. cooldowns.dev#javascript-e...
🚨 Active supply chain attacks on both PyPI and npm today. lightning 2.6.2/2.6.3 and intercom-client 7.0.4 are compromised with credential-stealing malware. Same attacker, same tooling, two ecosystems. Downgrade immediately.
Macil reposted
Yeah some unsolicated hot takes in no particular order 1. React is better than everything that came before it and I haven't seen anything do it better 2. Hooks are genius 3. If you've never done frontend on a moderately large app, you probably underestimate frontend engineering
it’s always funny to me when people try to paint react as some kind of “big company” propaganda. it was like one person’s wild idea and then there was at most ~5 people maintaining it for years. people HATED it when fb open sourced it. the only reason it won is because the model was actually good
Macil reposted
Macil reposted
Replying to @bruces.bsky.social
for what it's worth, the vulnerabilities described in the Mythos article that it found and exploited were of usual kinds. There are no mysterious move 37s. If AI discovers a new class of vulnerability that affects code patterns we strongly thought were safe then I'll panic.
When ChatGPT was new and everyone was anxious to understand the specific pace of AI progress, I figured a significant milestone to watch for was when it could find and exploit brand new software vulnerabilities on command. It's slightly unnerving to see the milestone met today.
New Claude found zero-day's in OpenBSD, ffmpeg, Linux and FreeBSD.
Macil reposted
"Over 99% of the vulnerabilities we’ve found have not yet been patched"... "Mythos Preview is capable of identifying and then exploiting zero-day vulnerabilities in every major operating system and every major web browser when directed by a user to do so ... oldest ... 27-year-old bug in OpenBSD"
This is the real deal folks red.anthropic.com/2026/mythos-...
Macil reposted
Macil reposted