🫠
Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications.
On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments.
While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software.
After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely.
openai.com/index/securing-ou…
+ use @DoxxnetHQ
Lots of Security incidents today, here's some practical tips for those who just wanna protect yourself:
1) UPDATE EVERYTHING - main targets are:
a) Browsers - Chrome, Safari, Firefox
b) OS: : Settings -> General -> software update; Linux: sudo apt update && sudo apt upgrade; Windows: I think there's a button. Do this for any remote home servers & Openclaw/hermes boxes
c) Phones: Update it all
d) Desktop apps (Slack, Cursor, Discord, etc)
2) Limit web browsing this week: Stay within walled gardens (twitter, youtube, etc); don't click random links this week. Just wait until the patches go through. Do step 1 again after a week.
3) Avoid untrusted Wifi networks until your devices are updated
4) Stay vigilant about phishing emails offering "certificate fixes" or fraud updates
5) Have an iPhone 11-16? TURN BLUETOOTH OFF for the time being until apple patches come through.
6) Code/Dev: UPDATE ALL YOUR REPOS - @charliermarsh has already updated uv 0.13.4 which reacts to the eMudra incident for uv (Python), Typesript according to Claude has equivelant. Tell your agents to go and update all your active repos to update package manager across uv, npm, bun, cargo, etc etc
What advice did I miss?
There are so many of these legacy software vendors out there running large parts of our physical infrastructure.
Scary to think about the surface area of such attacks due to legacy vendors’ lack of ability to innovate / buyer’s unwillingness to change the status quo (aka rip them out and replace them).
BREAKING: 🇫🇷 A major cyberattack has hit supermarket refrigeration systems across France, reportedly causing fresh and frozen food to spoil after hackers remotely manipulated temperature controls.
According to BFMTV, the attack began overnight from Sunday into Monday, targeting older versions of refrigeration-control software supplied by Danish company Danfoss.
The attackers allegedly exploited the systems’ defrosting function, forcing refrigeration equipment to heat continuously instead of keeping products cold. Temperatures were reportedly set as high as 50°C, destroying entire stocks of food within hours.
Several major supermarket chains have been affected, including Intermarché, Super U and E.Leclerc. Metro stores and independent businesses, including butcher shops also have been impacted.
Around 150 Intermarché stores are reportedly affected, according to industry sources cited by BFMTV. The full number of affected locations remains unclear.
The financial damage is already significant. In Haguenau, Alsace, an Intermarché store reportedly lost its entire frozen-food stock, worth approximately €50,000. A nearby Super U allegedly had to discard both fresh and frozen products, with losses estimated at €220,000. An E.Leclerc store in the same town was also affected, although its losses had not yet been calculated.
The combined losses reported at the Intermarché and Super U stores in Haguenau alone amount to around €270,000. Nationwide damage could reach several million euros.
Intermarché reportedly confirmed the incident in an internal communication, warning that several stores had suffered losses of frozen products. Coopérative U also acknowledged the attack and established crisis teams to assist affected stores and coordinate with refrigeration specialists.
Technical investigations are ongoing, and the full extent of the damage has yet to be established.
Authorities and companies are still working to determine the full circumstances of the attack. The identity of the perpetrators has not been confirmed.
We’re one of those firms but we didn’t hire devs to cut vendor costs (that’s a welcome side effect) - we hired them to train our own models, significantly increase our output, and punch above our weight.
Doing what we used to inside other tools a bit better is maybe 1% of the focus.
Talked to a finance guy today who said his firm has not only totally stopped buying software in favor of building it all in-house, they've actually hired multiple SWEs to do the buildouts, even though their salaries cost more than the subscription fees they were paying. Having tools custom-built for their specific business is just that valuable. So they're not spending less money or saving time, versus the pre-AI world, but they're getting a much higher return on the money and time they are spending. Much to ponder
In venture, the long game is the only game. Yet short-term incentives make it easy to forget.
Those who forget don’t last ...
Codex currently feels like Claude did a few months ago when everyone was using it:
Slow, poor quality, and not what it once was.
Pls get @sama more GPUs
Brilliant move to lock everyone into grok @bot + learn faster than everyone else.
Most founders get this.
But when the price at seed is 50% better vs seed funds, they don’t care.
(Plus who doesn’t think at seed that they’ll blow it out of the park and will have investors lined up to preemt their A so that all this doesn’t matter)
POV: multistage leads your seed round
> portfolio sick, network crazy, T1
> you imagine how dope it’s going to be telling ppl you raised your seed from [shiny name]
> launch tweet will get at least 2M views … you crack a smile
> they called you a “category defining generational company” at the closing dinner
> partner told you to get the lobster (MP)
> they backed Apple back in the day - maybe they see you as the next Jobs
> (memo: technical founders, cheap call option)
> brand kinda helps with recruiting
> but other investors see it as a call option, so not quite the flex you imagined
> lobster partner checks in quarterly
> Zoom background is a Starship launch. doesn’t agree with his politics, but n-of-1 visionary
> tells you he knows the CIO at SAP - can intro you to the Head of Infra at Oracle
> you remind him you’re building consumer hardware
> $1M run-rate, all in the last 8 weeks
> prep the Series A data room (DocSend with deck)
> then it dawns on you how much leverage [shiny name] has
> if they lead, makes sense
> if they follow on, why didn’t they lead?
> if they don’t follow on, uhhhh
> but I guess worst case it’ll still be cool to have “prev backed by [shiny name]” in bio
Every seed investment is a core investment where you put in everything you have to help make it successful!
Every seed investment is an option bet.
“You have to think about the venture portfolio as a bunch of options. Each seed investment is an option bet.
You are buying an option to see if it is an outlier, and you want to have enough of those to make sure you have an outlier.
When there is true quantitative evidence that it is going to be an outlier, then you position size up.” @venkyganesan
How do seed funds play in a world where mega funds are totally price insensitive and “option bets” is the mindset for seed checks @infoarbitrage @chudson @hunterwalk @honam @dunkhippo33 @jasonlk
I spent my entire weekend training our internal OS + sourcing engine / doing evals - not surprised Mercor, Micro1, etc are printing $$$ doing this for you
Training AI is the fastest growing job category in history.
In 2 years, we went from paying out nothing to > $6M / day.
Many of the top AI researchers say this will be the largest job category in the economy.
Learn more about it on 60 minutes!
How long until I get pitched by Griffin without realizing it?
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).
Readers added context they thought people might want to know
The 48% figure and "video Turing test" claim are from Tavus's own study of 54 one-minute calls, not independently verified or using a standard protocol. Griffin-Lite leads NVIDIA's VideoFDB benchmark on their public leaderboard.
cellcog.ai/blog/tavus-gri…
research.nvidia.com/labs/amri/proj…
tech-ish.com/2026/10/02/tav…
A must listen!
.@BarrettLyon built the first cloud DDoS-protection service, mapped the entire internet (now on display in New York's Museum of Modern Art), and got chased out of Moscow hunting Russian extortionists.
His new company is doxx.net, a parallel internet designed to keep advertisers, trackers, and creepy VPNs out of your life.
Barrett joins a16z's Joel de la Garza to explain what it is and how it works: a private network where messages, calls, and files travel directly between people instead of through someone else's server, and where nobody needs your name, number, or email to reach you.
They also get into why most VPNs are "just entry and exit points," why you can't set policy on equipment you don't own, the day Barrett's plane lost power over Oakland on the way home from a16z, and the data center myths that drive him crazy.
0:50 Meet Barrett Lyon
1:31 Losing an engine over Oakland
7:08 The hacker name behind the company
10:12 Censorship is coming to the US
13:17 Why VPNs barely protect you
14:21 Building a parallel internet
16:30 20GB files with no middleman
17:57 Free VPNs sell you
21:47 Why you must own the hardware
22:27 26 data centers, 12 people, run by AI
25:03 Data centers don't drink water
29:17 What ad trackers know about you
32:35 The DDoS attacks that built Prolexic
36:17 Mapping the internet for $50
39:46 Hunting extortionists in Moscow
41:47 Starlink at Burning Man
YouTube: youtu.be/bmDrbHOh7Bo
@BarrettLyon @YODELAGARZA
The internet was built to move packets, not to protect the people on it.
In practice that means someone else resolves your DNS, routes your traffic, and holds your messages, files, and call records. End-to-end encryption hides the content. The metadata (who talked to whom, when, from where) is still theirs.
(fun fact: there aren't any US based VPN / anonymizer companies anymore that don't use foreign-owned 3rd party infrastructure ...)
That was tolerable when it was one person leaving one trail. Now one agent reads your mail. One books your travel. One configures your infrastructure. Each acts as you, on the same public internet, at machine speed, leaving a trail that describes you better than you could yourself. The more we hand to agents, the more the internet knows about us.
It cuts the other way too. An agent with your credentials is one malicious page or one poisoned DNS answer away from working for someone else. Every agent you add is more attack surface.
Everyone is worried about what agents will say. Almost nobody is asking what network they run on, who can see them, and what happens when one gets compromised.
That is the problem doxx.net is built for:
They're pioneering Agentic Defined Networking: the most private and secure way for you and your agents to use the internet.
- Private, because your agent builds a network you own, with your own address space, DNS, certificate authority, and firewall policy, and then runs on it with you. Calls, messages, and files go device to device. No server in the middle. No email or phone number to sign up. Nothing to log.
- Secure, because threats are stopped before a connection exists. Malicious destinations are blocked at DNS, 38 million so far, and all of it runs on a backbone they built themselves rather than rented cloud.
I've had Doxx on 24/7 since I first got access. As of today, you can get access too
The open beta is live: point your agent at config.doxx.net and it will explain doxx.net to you and set it up.
PS: Thank you for letting us invest when you just got started @BarrettLyon and welcome to the team @a16 (@martin_casado & team)
Today we’re launching doxx.net, enabling Agentic Defined Networking for everyone. We raised $38M from @a16z, @AnimoVentures and Focal.vc.
Create private networks for you and your AI agents, without a server in the middle. 38M+ threats blocked since December.
Our first private-network call was a freedom call. Now it’s your turn.
Welcome @DoxxnetHQ!
I'm pretty sure it's been on 24/7 since I first started using it and I (+ my agents) won't use the internet without it on anymore!
Excited for the world to get their hands on what @BarrettLyon and team have been cooking up over the past few months!
Today we’re launching doxx.net, enabling Agentic Defined Networking for everyone. We raised $38M from @a16z, @AnimoVentures and Focal.vc.
Create private networks for you and your AI agents, without a server in the middle. 38M+ threats blocked since December.
Our first private-network call was a freedom call. Now it’s your turn.
Would have thought that meta focuses on consumers and smb for now but this move makes sense.
Another very potent entrant in the “ai for enterprise” arena. This one will be fun to watch.
We believe superintelligence will create significant new opportunities for all people and businesses. Meta already serves billions of people at scale and helps hundreds of millions of businesses reach customers. Today we are starting the next major pillar of our business, Meta Enterprise Platform, to help businesses use AI to grow and transform in new ways as well.
A $3B gift to CMU and Miami.
Another huge win for the most exciting city in the western world.
While everyone else is caught up in 0 sum thinking & increasing taxes (SF & NY 👀), Miami continues to grow the pie.
Welcome to the neighborhood @CarnegieMellon
Today, Carnegie Mellon University and Ken Griffin, founder and CEO of Citadel and founder of Griffin Catalyst, announced a historic $3 billion gift to elevate Carnegie Mellon in Pittsburgh and launch Carnegie Mellon University Miami. #CMUMiami
Learn more: cmu.edu/news/stories/archive…
Last week, we hosted the first Miami Founder Summit together with @thelabmiami.
60 Miami-based, venture backed founders together in a room for an afternoon for amazing sessions and conversations led by @ClemDelangue, @nejatian & @BarrettLyon.
It's amazing to see how many great people have moved / continue to move here. Now we just need them to update their LinkedIn to Miami too 😉.
Thank you to those who came and to @ayalstern and the team at The LAB for being such great partners on this.
@focal_vc
