Pinned Tweet
CYBER NIGHTMARE | DEEPSEEK RAISES $15B | BLAKE SCHOLL ON MTS xitter.cf/i/broadcasts/1DGleVyRz…
Ex-OpenAI science head @kevinweil explains why AI is making breakthroughs in math and coding but still struggles to discover new medicines:
"We now live in a world where it doesn’t matter whether you learned how to program, you can turn any idea you have into an actual working product in a few minutes."
"You got models solving Navier-Stokes and making progress on the Riemann hypothesis. But I think the biggest impact that AI will have for the world will be in sciences."
"These fields of science are so empirical. You don’t just think really hard about how you find a molecule that binds to the protein. You run experiments. If we want models to do that, the models need to see that data at scale."
"This is why there’s such a difference between model performance in mathematics and software engineering and model performance in science."
@warpsciences
On #1: I was previously CPO and VP of Science at @OpenAI, and I couldn’t be more optimistic about the future. Anyone can use AI today to build an idea in minutes, and we’ve just seen models solve Navier-Stokes and make progress on the Riemann Hypothesis.
But the reality is that AI is not yet at the same level in the physical sciences, especially in places like drug discovery where it could do the greatest good.
The reason is simple: these are highly emergent and empirical sciences, and they move forward through data and experiment. The models don’t have access to the type and volume of experimental data they need.
Ex-OpenAI science head @kevinweil reveals how America outsourced a critical 3-5-year stage of drug development entirely to China:
"Imagine you and I start a drug discovery company. You have a hypothesis that if we could inhibit this protein with a drug, we have a positive effect on people’s cancer."
"You find a molecule that binds to the protein, but most likely that first molecule is gonna be toxic, or not soluble, or not very potent. Instead of a little pill, you have to take a kilogram of it."
"You spend the next three to five years making variations of this molecule hoping that you find one that is less toxic, more potent, more soluble."
"The thing that people don’t know is that whole three to five years of making and testing molecules, we’ve outsourced it entirely to China. We cannot depend on a geopolitical adversary to do this work for us."
@warpsciences
💥 Introducing Warp Sciences, a new company applying AI and robotics to help the world develop new medicines faster.
We’ve raised a $215M Series A led by @lightspeedvp to bring frontier AI out of the browser and into the lab, and build scientific capacity here in America.
I co-founded Warp with my long-time friend @dpatil, FAIR co-founder @manohar_paluri, and drug hunter Sol Reisberg. The early team is full of incredible people like @bkspears9, who helped achieve Q>1 fusion at LLNL.
Ex-OpenAI science head @kevinweil on his $215M plan to build robotic labs that run 24/7 and bring drug research back from China:
"We’re building robotic labs that do science at scale, and we’re doing it here in America. Doing science at scale is the way we generate the training data the models desperately need."
"We need to reshore scientific capabilities from China so that we don’t lose our pharma industry the way that we lost solar and drones and manufacturing."
"We’re not developing drugs on our own, we are building the infrastructure so that everybody who does develop drugs can go faster."
@warpsciences
💥 Introducing Warp Sciences, a new company applying AI and robotics to help the world develop new medicines faster.
We’ve raised a $215M Series A led by @lightspeedvp to bring frontier AI out of the browser and into the lab, and build scientific capacity here in America.
I co-founded Warp with my long-time friend @dpatil, FAIR co-founder @manohar_paluri, and drug hunter Sol Reisberg. The early team is full of incredible people like @bkspears9, who helped achieve Q>1 fusion at LLNL.
Project Eleven CEO @apruden08 warns AI will put everything exposed to the internet under constant attack, just as crypto has been for years:
"No one cares about security until they do. What people are coming to appreciate about security in the AI era is that everything that’s exposed to the web will be under constant attack."
"The cost of attacking this infrastructure, if you’re a hacker, has dropped to basically zero. You can spin up any amount of agents and start pressure testing infrastructure all the way down to open source libraries."
"This is similar to how crypto has been from the beginning. I’m trying to think of a word that’s nicer than paranoid, but you kind of need to be paranoid. In the post-AI cybersecurity landscape, everyone is going to need to have a security-first mindset."
@projecteleven
Assumes three things:
(1) that the state of the art models are widely available/open-source. That is contrary to what my expectations are w.r.t. incentives. Whether for safety or revenue, OpenAI and Anthropic have reasons to gate their best cyber models and make you pay up for an AI-audit basically.
That doesn't slow down sophisticated attackers much: nation-states will pay or build their own anyway. What it does is make AI-assisted defense something only large organizations can afford, leaving the long tail, where most of the easy targets are, exposed. (goes without saying that open source libraries maintained by volunteers are probably toast no matter what)
(2) that the security vulnerabilities themselves are "shallow" and easily detectable. If bugs are deep, finding them takes a lot of tokens per line of code, and that's where the asymmetry bites: because defenders have to secure the whole surface area of their codebase against attack, dollar for dollar they will not be able to spend as much in tokens to *harden* every LOC as the attackers will be able to spend to *break* one specific point.
A few reasons to doubt (2):
a. The PHP vulnerability was a very deep bug that many people overlooked.
b. It's been shown that just spending more tokens can get better results. (e.g. the Claude Riemann-Zeta result, where the engineer basically just kept prompting the model to "keep trying")
c. ecdsa dot fail, the more tokens get thrown at the problem, the more progress gets made. Implies that spending more on tokens surfaces more insights as you get deeper into the problem.
On their own, b. and c. cut both ways: if more tokens produce better results, defenders get more tokens too. They favor the attacker because of the asymmetry: the defender has to spend tokens hardening every line, while the attacker only has to spend them on one.
(3) that the pace of software development slows down to the pace of security engineering. If anything, AI is vastly *accelerating* the product/engineering cycles, and exponentially *more code* will be produced in the coming years. Hard to imagine that security tools will be able to keep pace (security is always ignored by everyone until it's impossible not to).
Defenders do have home-field advantages: source access, and the chance to scan code before it ships. But that only helps the orgs that actually do it, which brings us back to (1) and (3).
So it's not obvious to me that AI favors the defense. In fact, at the limit things will settle to roughly where they stand today: attacker has slight advantage based on the asymmetry of resources they can bring to bear against one, carefully selected target.
AI tools can certainly help improve security, but not enough to overcome this fundamental asymmetry and flip the advantage to the defender, imho.
SITUATION EXPLAINED: DeepSeek is closing a $15 billion raise, and its backers include Tencent, CATL, and a dishwasher maker.
• The round is about 100B yuan, double the original 50B yuan target, at roughly 500B yuan (about $70B) before the new money
• Carmaker Geely, delivery giant Meituan, and chipmaker GigaDevice joined too
• Most of the backers are companies in China’s tech supply chain rather than VC firms, which could make DeepSeek a core part of that chain
• That valuation is a fraction of what top US labs are worth, with Anthropic’s last round valuing it at $965B
• DeepSeek hired CITIC Securities in September to prepare a possible IPO on Shanghai’s STAR Market
@theojaffee: "It’s a sort of flywheel where DeepSeek helps Huawei build chips that are then used to train the next DeepSeek model that then makes DeepSeek money that they can use to invest in chips made by Huawei. It’s a virtuous cycle."
SITUATION EXPLAINED: Elon Musk just backed a $3.5 trillion merger of SpaceX and Tesla, which would be the biggest deal in history.
• Musk backed it on an investor call, but he’s conflicted and has to stay out of the talks, so insiders worry it’ll draw shareholder scrutiny and add weeks of delay
• Independent committees at both companies are weighing SpaceX buying Tesla or a new holding company, likely in an all-stock deal
• It echoes Tesla’s 2016 SolarCity buyout, where Musk was also on both sides, which led to years of lawsuits
• Tesla shareholders would likely want a premium, which Wedbush’s Dan Ives puts at about $500–550 a share
• Musk has said since 2024 he wants about 25% voting control of Tesla, and analysts say a merger would give him control of the combined company
• The companies’ AI and chip work now overlaps, and closing before Trump’s term ends could help the deal avoid a tougher antitrust review
@theojaffee: "Elon has just been grinding away at the exponential for years and years and years."
Kong CEO @sonicaghi warns AI is giving hackers nation-state-level capabilities while companies deploy agents they can't secure:
"In the last 24 hours we had PHP2Shell, a bunch of gateways got compromised, and the frequency’s increasing. You have now nationwide type of attacker powers by leveraging models, and the companies haven’t defended themselves through security models."
"Agentic enterprise and agentic coding is creating a whole new surface that most of the time is non-deterministic. That creates a lot of empty spots to be attacked."
"We do rate limiting, blocking, and a kill switch. This is becoming a large risk into boardrooms of the Fortune 500s."
@kong
Two critical vulnerabilities have been identified in PHP, a widely used programming language that powers websites and web applications. These vulnerabilities (CVE-2026-109618 and CVE-2026-109619) could allow malicious cyber actors to remotely execute code in PHP core, potentially compromising sensitive data and networks.
The FBI urges organizations to take the following steps now:
➡️ If an immediate update is not feasible it is recommended that organizations take the system offline until it can be fully mitigated.
➡️ Immediately update affected PHP installations following the guidance on php.net. Inventory where PHP runs in your environment, prioritizing internet-facing systems and vendor products that bundle PHP.
➡️ Monitor web servers for suspicious activity, including PHP processes unexpectedly launching command shells or downloading files. Ensure appropriate process-monitoring capabilities are in place to help identify potential exploitation.
➡️ If you find signs of compromise, isolate affected systems, preserve your web server and PHP logs, and report it to your local FBI field office or ic3.gov. Those reports help us identify the actors involved and warn other organizations before they become victims.
For the latest information, follow PHP’s announcement on php.net, CERT-EU Security Advisory 2026-018, and CISA’s Known Exploited Vulnerabilities catalog.
ALT Critical vulnerability alert for PHP Core with CVE-2026-109618 and CVE-2026-109619, CVSS score 9.2, urging immediate patch and threat hunting.
Kong CEO @sonicaghi reveals token traffic through its gateways is growing 140x year over year, approaching 1 quadrillion tokens:
"The number of tokens is growing exponentially. For the subset that we see, we are running about 500 trillion tokens, going to a quadrillion soon, because it’s growing 140x year over year."
"The cost per token is going down, and inference will still be exploding. We’re gonna live in this enterprise world that is more multi-model, cheaper tokens, cheaper levels of intelligence."
"The next pieces are, how do we secure it? How do we govern it? How do we cost control it?"
@kong
Former DOE security engineer @asynchronous_x explains why America's critical infrastructure hasn't been attacked more often despite its networks being riddled with security holes:
"The work I used to do for the Department of Energy was around critical infrastructure. They’ve relied on obscurity, you don’t know that they’re there, so why would anyone attack us. As soon as you took any sort of magnifying glass to their networks, they were usually riddled with holes."
"The reason they haven’t gotten attacked so much is because the repercussion is so much higher. If you compromise a water treatment facility and you accidentally end up killing people, that’s gonna have extreme repercussions versus if you steal some cryptocurrency."
"We’ve been relying on political and other methods for keeping them safe rather than actually making them secure."
Two critical vulnerabilities have been identified in PHP, a widely used programming language that powers websites and web applications. These vulnerabilities (CVE-2026-109618 and CVE-2026-109619) could allow malicious cyber actors to remotely execute code in PHP core, potentially compromising sensitive data and networks.
The FBI urges organizations to take the following steps now:
➡️ If an immediate update is not feasible it is recommended that organizations take the system offline until it can be fully mitigated.
➡️ Immediately update affected PHP installations following the guidance on php.net. Inventory where PHP runs in your environment, prioritizing internet-facing systems and vendor products that bundle PHP.
➡️ Monitor web servers for suspicious activity, including PHP processes unexpectedly launching command shells or downloading files. Ensure appropriate process-monitoring capabilities are in place to help identify potential exploitation.
➡️ If you find signs of compromise, isolate affected systems, preserve your web server and PHP logs, and report it to your local FBI field office or ic3.gov. Those reports help us identify the actors involved and warn other organizations before they become victims.
For the latest information, follow PHP’s announcement on php.net, CERT-EU Security Advisory 2026-018, and CISA’s Known Exploited Vulnerabilities catalog.
ALT Critical vulnerability alert for PHP Core with CVE-2026-109618 and CVE-2026-109619, CVSS score 9.2, urging immediate patch and threat hunting.
Former DOE security engineer @asynchronous_x warns AI is narrowing the gap between junior hackers and agencies like the NSA:
"We used to have this thing in industry where we would say, this is pretty secure, unless your threat actor is the NSA."
"Now with AI tools you can take more junior people with less expertise, and they can make more sophisticated attacks. Somebody that would’ve been a less high-level threat actor before could potentially become a much more dangerous threat actor."
"There’s plenty in the industry that think we should limit access to these tools because it can make smaller threat actors much more dangerous."
Some stats on #php2shell. In the 10 hours before it was publicly announced, across the @Cloudflare network we saw thousands of attempts to exploit it. That jumped to tens of thousands in the 45 minutes post announcement, hundreds of thousands within 2 hours, and millions now.
Former DOE security engineer @asynchronous_x explains how a miscommunication turned a planned PHP2Shell fix into a race against hackers:
"It should have been the best-case scenario. Security researchers identified these vulnerabilities, developed proof of concept exploits in advance, and were informing the PHP core team for fixes. That whole part was working fine."
"The reason this one’s unique is there was basically a leak of the vulnerability before the disclosure. That was supposed to be released on October 22nd, today. Unfortunately, there was a miscommunication, and they released some of the details early. So there was this panic race where the problem is out in the wild, but we don’t have a fix prepared yet."
"The thing it’s taught the security realm is that we probably can’t rely on coordinated disclosure anymore. If the speed at which people can develop exploits is now so fast that we can’t possibly rely on things not getting leaked, then it’s not something we can lean on as much as we used to."
OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites.
Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors.
Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact.
This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare.
openai.com/index/coordinated…
Readers added context they thought people might want to know
Wiz Research wasn't the first one to discover the exploit. An independent researcher first discovered it: x.com/calif_io/statu…
Synthesia CEO @vriparbelli reveals how LLMs have become surprisingly good at video editing, making it possible to create entire videos through chat:
"You have to write a script. LLMs are pretty good at that. You have to generate the actual assets, that’s usually what you capture with a camera."
"But then there’s a process after that where someone takes those assets, takes the script, and edits everything together."
"The big thing that’s happened in the last couple of months is that LLMs have now become an amazing video editor as well. All the motion graphics, text on screen, how to delineate the different scenes."
"I don’t think anyone really expected we’d get this good this fast. We’ve gotten very, very close to what it would be like to work with a really good video editor."
@synthesiaIO
Syren Video is now live and FREE to try!
ChatGPT moment for agentic video.
Prompt → agency-quality AI video → chat to edit.
Powered by Opus 5.5 + Syren renderer w/ 3D support.
Browser or Claude MCP.
Insane examples in thread 🤯🤯🤯
Try it free here: shorturl.at/r57qy
.@deredleritt3r explains why OpenAI's latest math breakthroughs have him convinced AI could be on the brink of RSI:
"When I used GPT-6 Astra, I messed around with a couple of ciphers that were previously unresolved, from World War I, and I successfully had Astra decrypt a couple of them."
"But every one of the results was the model having the key, and brute forcing the solution by searching through the entire space of possibilities. I have not been able to get Astra to come up with the key on its own based on some novel idea."
"The math results recently achieved by OpenAI’s internal model have really updated me. I think Terence Tao said about the quasi-Riemann solution that it was a combination of three or four existing ideas but arranged in a very unusual way."
"I’m guessing an internal model at OpenAI is probably still not able to achieve RSI. But it feels like we’re really not very far away. Once the models are able to do something, in 12 months or sooner, they’ll be really, really good at it. I think we’re on the brink."
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math
.@deredleritt3r warns the emerging cyber crisis feels like February 2020, when nobody knew how bad COVID would get:
"This reminds me of February 2020, when we heard about COVID-19 for the first time, and everyone was like, how bad is it gonna be?"
"The US federal government, because they’ve had access to Mythos since basically February of this year, I am hoping and praying that they have, secretly from the public, strengthened a lot of the really critical systems. But we can’t know for sure."
"Right now anything that is vulnerable is going to be attacked. Two-factor authentication, change your passwords. Do as much as you can."
.@deredleritt3r argues real RSI begins when AI can make Ilya Sutskever-level breakthroughs on its own:
"It’s a good thing that you’re talking to a lawyer because definitions is what we do. I focus on the word self. You need to be able to train a model that is your successor and make it more capable than you are. Until we have that, I don’t think it’s RSI."
"Right now we don’t have RSI, and I don’t think that the labs have it internally either."
"We will have it when a model is capable of fully autonomously coming up with really creative ways to make the new model more capable. Ilya Sutskever type brilliant moments. I think that’s coming pretty soon."
.@deredleritt3r warns the cyber apocalypse may have already begun:
"When you have an event followed within 30 minutes by another event, followed in an hour by another event, and so on every couple of hours, it starts feeling like we’ve crossed some kind of threshold."
"This has been expected by many. I wouldn’t be surprised if it’s happening now. It certainly seems to be happening now."
SITUATION EXPLAINED: An ex-OpenAI researcher is asking whether "internet lockdowns" are coming as AI-powered hacks pile up.
• Shopify's @tobi called it the "craziest day in cybersecurity in a long time," and another researcher worried it's "the early stages of a global cybersecurity incident"
• Former OpenAI chief futurist Joshua Achiam says the security holes were always there, AI just makes them easier to find, and defenders have more money to fix them
• In a "moderately bad" case, he sees stock markets closing for a few days, or China using the chaos to move on Taiwan
• To stay ahead, he says defenders need more compute than attackers, maybe through "national compute reserves allocated to cyber defense"
• Only about 4 of this week’s ~16 big incidents actually happened this week. The rest were found earlier and disclosed now
• AI researcher John Allard says this is what it feels like when AI capabilities "stop being hypothetical," and he thinks 3D, animation, and video editing are next
@theojaffee: "The worst case scenario would look something like literally your power goes out. It would look something like the water in your tap stops running. It would look something like you go on the internet and everything is gone, and none of this was ever disclosed in advance. And what we see now is not that."
SITUATION EXPLAINED: It’s being called "Black October," the worst week in cybersecurity in years, with AI on both sides of the fight.
• PHP2Shell, a hole in the software behind about 70% of websites, has been patched, but attackers have already tried to exploit it millions of times
• eMudhra, an Indian company that vouches for websites, made its private keys so weak that one was literally the number 4, so anyone could fake a secure HTTPS site. Chrome, Firefox, and Apple have since blocked them
• GitLab had a flaw rated a perfect 10/10 for severity: "anyone on the internet can fully take over the affected system easily and silently"
• LiteLLM, a popular AI tool, briefly shipped an update with a password stealer hidden inside
• Hackers also stole data on millions of California students, knocked grocery giant Ahold Delhaize offline, and hit AI DevOps startup Zeabur so hard it’s shutting down
• For now, you’re much safer on big platforms like Google, Apple, and Meta than on a small WordPress site that hasn’t been updated in years
@theojaffee: "Some of these are just, like, really poorly secured startups that probably vibe coded all of their code. Others are vulnerabilities that existed for many years without ever being exploited or even discovered until they were discovered by an LLM, which is what happened with PHP."
This is the most serious 48 hours in cybersecurity history.
"Black October" they're calling it
It seems every *hour* there's a new incident, that in a typical year, would be one of the biggest incidents of the *entire year*
SUMMARY OF THE LAST 48 HOURS:
YEAR-LEVEL INCIDENTS (In a typical year, these would be among the biggest incidents of the year):
1. PHP2Shell: a hole in the software behind ~70% of websites. Governments took sites offline.
2. eMudhra: the master keys that prove websites are real were cracked. One was the number 4.
3. California: data on every public school student since 2023 stolen, including disability status
4. Railway: hackers stole customers' secret keys from a hosting platform used by millions of developers
5. Zeabur: 321,810 accounts breached. The company is shutting down.
6. LiteLLM: an AI tool downloaded 3.4 million times a day was poisoned with a password stealer
7. GitLab: a 10-out-of-10 severity hole in every version since 2014
8. Cairn: criminals used an AI hacking tool to break into 100+ companies
9. Ahold Delhaize: hackers knocked the owner of Food Lion and Stop & Shop offline
10. Onamae: malware planted on up to 26,739 customer servers in Japan
MONTH-LEVEL INCIDENTS (In a typical month, these would be among the biggest incidents):
11. Signal: a bug could leave your mic and camera on after you hang up
12. NGINX: 3 critical holes in software running ~1/3 of websites
13. HTTP/3: a single connection can crash 7 of the 10 web server types tested
14. Plex: hackers breaking into home media servers
15. SafePay ransomware: FBI warns it has hit 600+ critical infrastructure organizations
16. Palo Alto: ransomware gangs breaking in through company firewalls
17. Bullhorn: platform used by 10,000+ companies knocked offline
18. Medi-Cal: 970,000 patients' data stolen
19. WordPress: hackers stole a plugin maker's keys and used them to plant backdoors on websites
20. AOMEI: backup software's update server handed out a backdoor
21. Meta Muse: Meta's AI agent exposed users' personal and business secrets
22. Tor: 333 relays hijacked to steal logins
23. TiDB Cloud: database company's cloud breached
Also serious:
24. State Bank of India: India's biggest bank confirms a data breach
25. OVHcloud: hackers slipped a password stealer into a cloud company's official tool
26. Siemens: a software hole being used to spread malware
27. North Korea: state hackers launch a new attack campaign on South Korea
