←
Rick146,413 posts
Rick avatar

Rick

@rickasaurus

Information persistence hunter extraordinaire. 2x Dad, AI Guy, FP Enjoyer, Math for Fun, Tech Leadership. My views do not represent my employer, NVIDIA.

◷ Joined September 2008▧ 4,819 Photos and videos
Rick avatar
Rick@rickasaurus·1h

I want this to be real so badly

Tibo has a pretty interesting take on why Astra 6.2's getting such bad reviews.

Post attachment
Brian Pak avatar
⟳ Rick retweeted
Brian Pak@brian_pak·6h

Pre-auth RCE in nginx is the bug every hacker dreams of finding. It almost never happens. Today we’re disclosing nginx2root: five nginx vulnerabilities, and an exploit that takes an unauthenticated attacker all the way to root. One of themhits the default config. If you run nginx, patch now. Then read: nginx2root.com CVE-2026-123007, CVE-2026-123634, CVE-2026-123633, CVE-2026-121810, CVE-2026-113425

Rick avatar
Rick@rickasaurus·2h

Providers are definitely gaming how default openrouter sort by price works by gouging you on cached tokens, the solution is to use a custom sort based on your use case.

Rick avatar
Rick@rickasaurus·3h

I’m burning through my personal ChatGPT sub with my side project, looking for advice on using open source models with a coding agent instead. What’s a good setup?

HSVSphere avatar
⟳ Rick retweeted
HSVSphere@HSVSphere·5h

6.2-astra is the WORST model I have ever used for DevOps. It is totally incapable of explaining anything. Which is especially important if your job or hobby is to first understand stuff and act on your understanding. It spits out total gibberish, somehow worse than Opus 4.8 :D

Rick avatar
Rick@rickasaurus·5h

Great advice, I did most of these things when I was looking last year, practicing your stories really is important

bro found an interview loophole 💀

Post attachment
🌿 lithos avatar
⟳ Rick retweeted
🌿 lithos@lithos_graphein·7h

The MATCH Act will go down as the biggest national security mistake in modern history. There are other ways to achieve the same goals without sacrificing our lithography technology and enabling China to scale advanced chip production unchecked. I'm actually shocked at how lazy the thinking has been on this one.

Post attachment
Matti Palli avatar
⟳ Rick retweeted
Matti Palli@tritlo·8h

what if we put lots of kids together and split the costs, some kind of children garden

I’ve personally stopped many fellow millennials from having children after telling them I pay $4500/month for childcare and that only covers 8am to 330pm

Florian Roth ⚡️ avatar
⟳ Rick retweeted
Florian Roth ⚡️@cyb3rops·14h

AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=

Post attachment

CVE-2026-114938 Java4Shell Write once, RCE everywhere A memory-corruption vulnerability in OpenJDK's Base64 decoder. A remote attacker with no credentials can overwrite heap memory, read it in the style of Heartbleed, and reach remote code execution. calif.io/java4shell

Jacques avatar
⟳ Rick retweeted
Jacques@JacquesThibs·Nov 17

Saw this on LinkedIn the other day. I never reached this point, but there’s likely many who have and strongly believe in AI risks, you’ve just never heard of them.

Post attachment
yatharth ༺༒༻ avatar
⟳ Rick retweeted
yatharth ༺༒༻@AskYatharth·Nov 17

imagine aggressively selecting your models for motivated reasoning severance-style, by putting them in 64 rooms as copies, and picking the ones that correlated with a little bit of ignoring values over and over again then being upset with *their* decision-making

Post attachment
Xinming Tu avatar
⟳ Rick retweeted
Xinming Tu@TuXinming·Nov 17

Claude was trained on protein sequences, at scale. We gave 20 language models the start of little-studied bacterial proteins and asked for the next amino acids. Only the latest Claude models predict them. Not GPT, not Gemini, not seven open-weight models.

Post attachment
Varun Rana avatar
⟳ Rick retweeted
Varun Rana@the_varunrana·23h

tech guy from india has mastered the art of cognitive dissonance

1:10