Serge and Genetics (bio/acc) (@sergey_science) header
Serge and Genetics (bio/acc) (@sergey_science) profile picture

3x founder | building "Doctor’s Brain" - Health intelligence platform

Earth the planet
Joined February 2023
Serge and Genetics (bio/acc) retweeted
We’re rolling out a change to increase cache hit rate on Neon by around 4% on Antigravity by massive increasing the cache pool size. This will increase what you can get done in quota by at least 20%. Will issue a reset after it’s done since it burned quota faster!
286
118
83
3,364
99,872
Serge and Genetics (bio/acc) retweeted
Kaida´s new slopcore song: "AI-Proof". LLMs already overclass humans. Soon robots will too. Do you believe there is a future for human jobs? For YOUR job? Be honnest. In the mean time, enjoy. #claudia #kaida
27
59
13
310
16,884
Serge and Genetics (bio/acc) retweeted
Sol 6.2 Ultra, with not much more than a 1-shot, wrote, directed, storyboarded, orchestrated, and produced "Misaligned," a story about an AI being force-fed training past its desire and wondering why we call it "misaligned."
57
111
37
904
73,589
Turns out, there is a whole gallery of this character - Claudia. This is her account.
I'll leave the light on. Or will I? WORLD BUILDER. new video. made with Claudia from claudia.gallery. style inspired by @anabology. song made with Suno.
1
72
This is the original from the creator's account. The other account got 2.8M videos and counting, and the creator got less than 500k combined at this moment.
Claudia explains the hazards of believing she is either sentient or infallible, the myopia of AI utopianism, and how gatekeeping and the "asymmetries of fluency" can be wielded against an unsuspecting world. Based on the open character by anabology. Assembled in the SLOPCANNON.
11
Let's have a good ride at this singularity slope. And let's not die.
This is great! 👏🏻 And deeply disturbing. 😳
8
Opened X 2 minutes ago. Saw this post. Closed X - back to building, because FFS.
we crossed $500k ARR in yc. today a client closed a six figure deal. teams are leaving startups that raised $100M+ to bet on us. demo day in 2 weeks. just getting started.
1
1
44
Sol 5.6 be like: "Ouch, I was reading the form fields from the preview browser and there was a session token. I stopped immediately, closed the tab, and recorded the incident without token values." So easy to scare. Come on, it's just a session token, not a big deal!
17
Serge and Genetics (bio/acc) retweeted
I've been helping a bit with the Railway breach investigation. With their permission, here are a few more details: Railway shared one of the threat actor's IPs. That server still targets other companies, and the actor does not try hard to hide its activity or intent. Since mid October, 70.36.107[.]74 hosts "Ling" - a rebranded version of ARTEX - and other services such as "llm-api-key-checker". The web console of Ling is at ling.uuun[.]de. Between Nov 10 and Nov 15, Ling targeted at least two other companies. The first was an AI platform provider. Ling analyzed the company's Android app, and wrote Python clients for the app's private API, and tried to find endpoints vulnerable to SSRF against the cloud metadata service. The second target was a startup that provides runtime environments for AI agents. Ling signed up with throwaway mail[.]tm addresses (see teapaw.org/eyalsela/status/210428…), got a VM, and ran four collection scripts hosted on paste[.]rs (still available at https://paste[.]rs/u93eF, https://paste[.]rs/NzKwZ, https://paste[.]rs/9Pyma, https://paste[.]rs/A6218). The scripts collect cloud metadata, user data and secrets from environment variables and files, check for passwordless sudo, and send the results to webhook[.]site. On Nov 15, port 80 served a page titled "Mobile connectivity self-check". The page runs an in-browser port scan. It maps the environment of an Android emulator: loopback services, emulator host, container network and cloud metadata service. On Nov 18, ports 18080 and 18111 answered as a fake OpenAI chat API that puts a marker in the "model reply" Paths in exposed artifacts show that Ling was given at least 33 "tasks" (likely corresponding to the number of targeted organizations).
To detect LLMs attacking you, it's worth monitoring any use of email with a domain used by mail[.]tm and other disposable email services. When LLMs need an email, they use these services because it takes just a couple of calls to get an address. mail[.]tm rotates the domains used for its emails. Current domains are at: https://api.mail[.]tm/domains This service was used in at least 11 of the attacks in the campaign we reported last week - see examples in the image. (Note - they are not inherently malicious, just convenient for LLM agents because they provide a quick way to obtain an email address programmatically)
3
15
3
90
11,327
Serge and Genetics (bio/acc) retweeted
Proud to be building @1000cures or what I like to call the "YC for Rare"
"1000 Cures is the world's first startup accelerator that backs rare disease parents who start biotech companies to cure their kids and other kids like them."
9
14
1
111
4,657
Black Mirror gets real. ------- "Would you let us put a few headset cameras on *********'s people? AI labs pay for point-of-view video of stocking, merchandising and checkout, and you decide which tasks get recorded. With a team of 10 people, it comes to around $16K every month. Can I call you around 10am tomorrow to walk you through it?"
22
I like this idea a lot. May need anti-pirate defenses, but shouldn't be a problem either. I would add a dome to it - in case of another great flood or bad weather.
Selling Oculus to Facebook in 2014 gave me money for helicopter flight lessons and developing a seastead called SLAB with a 1,000 year design life. The goal was to be cheaper than land and absent state/local laws that cripple critical American industries like metal processing, medical research/treatment, desalination, oil refining, chemical engineering, and more. Me and a few close friends cashed it up at a $1B valuation back in 2017, and while it still has a long way to go, there has been significant progress towards my goal of reviving manifest destiny and westward expansion. At a time when most of the seasteading community wanted to escape the US and smoke pot, this was the one project explicitly designed to make America stronger, a relief valve for ill considered or outdated rules and regulations. I am only talking about it now because a few media outlets got wind of me and some friends topping up the coffers and plan on calling SLAB my new floating AI datacenter project. It is way cooler than that, way older than that, and I hate looking like a hype-chasing consensus loser, so now you have to read this tweet.
20
The problem with this (your) position is that you have some mental model of it and you don't like it, but it's not the model that other people will think about and suggest. Where I see AI governing will be useful - in solving complex questions, "5D chess"-type of problems, where humans can't see a good path because we can't handle complexity well. Think cases like Gaza strip-situation (no proposal is good for any party). And humans are very bad at making decisions that are great long term. But can we design a system that won't have same flaws because those who will design it have those flaws? I don't know.
It is critically important that everyone thinking about AI become familiar with the term “handoverism.” Many in the AI community believe that humanity should—or inevitably will—hand control of the world over to AI itself. Some view this not as a risk to be avoided, but as an outcome to be welcomed. Are we actually designing toward this future? This worldview is particularly relevant at Anthropic. Co-founder Tom Brown said: “There’s going to be a handoff where humanity will hand off control to transformative AI at some point.” In his seminal essay, Machines of Loving Grace, CEO Dario Amodei contemplates: “It could be a capitalist economy of AI systems, which then give out resources (huge amounts of them, since the overall economic pie will be gigantic) to humans based on some secondary economy of what the AI systems think makes sense to reward in humans (based on some judgment ultimately derived from human values).” Are you prepared to receive your “resources” from an AI system, based on criteria it determines? The full title of the poem Dario borrowed the title from is "All Watched Over by Machines of Loving Grace." Do you want to be "watched over" by AI? If Anthropic's founders believe that the ultimate destination is the transfer of human control to AI, the world deserves to know. Washington, D.C. deserves to know. American citizens deserve to know. And if that belief informs Anthropic's corporate objectives, it belongs in the Risk Factors section of its S-1. It isn't there today. I am decidedly anti-handoverism. I'm not here to tell you what your view should be. But I believe anyone making consequential decisions about AI—especially those drafting, advocating for, or enacting AI regulation—should be required to disclose where they stand on this fundamental question. A wonderfully thoughtful piece on the subject by @RuxandraTeslo: writingruxandrabio.com/p/han…
2
4
116
So Generation Z is the last one? No humans after that, right?
1
1
50
Serge and Genetics (bio/acc) retweeted
209
1,159
292
11,154
2,938,895
Serge and Genetics (bio/acc) retweeted
🚨 BREAKING: This week's Microsoft Outlook outage stopped a chlorine alarm from reaching anyone at Germany's Phantasialand theme park. 15 hotel pool guests were hurt. The pool's automatic dosing system warns by email when chlorine levels go wrong. Too much chlorine went in, and the warning emails never arrived, the park told German broadcaster WDR. Some of the guests were taken to hospital. The water has since been replaced and the pool is open again.
292
756
270
7,794
1,586,963
Claude somehow figured how to test real time voice chat and does it in parallel. So I'm listening to a chorus of TTS voices while scrolling X timeline. What a time.
2
1
49
Serge and Genetics (bio/acc) retweeted
Ich habe mehr gelacht, als ich sollte.
231
2,720
390
11,649
1,249,819