Building at OpenAI. Prev @GoogleDeepMind @heroku
Mountain View, CA
Joined March 2013
- Tweets43K
- Following3.4K
- Followers5.3K
- Likes605
Dominic Kramer retweeted
During this red-teaming exercise, we also found a vulnerability that allows for unauthenticated remote code execution in Brotli, a widely-deployed compression library included in many pieces of software such as Node.js, Bun, and Envoy.
The flaw could let an attacker run their own code on affected systems by sending or replying with malicious compressed data, without any need to create a user account. That is, an unpatched server that accepts compressed input could be taken over by a malicious client, and an unpatched client that accepts compressed input (such as “curl --compressed”) could be taken over by a malicious server.
This root cause issue was actually fixed in the Brotli source code in February, but it wasn’t known at the time that it gave rise to a security vulnerability and there was no more recent release of Brotli. Since Brotli is built into other software, any software which includes this library needs to take the update too, followed by the users of that software. We disclosed the vulnerability to Brotli and other downstream projects such as Node.js, which released or are currently releasing security updates, and Brotli cut new release versions and published a CVE.
More details on the patch and mitigations below:
patchtheplanet.org/brotli
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…
a fix that sits in source with no release is basically no fix for most users. they update from releases, not commits
During this red-teaming exercise, we also found a vulnerability that allows for unauthenticated remote code execution in Brotli, a widely-deployed compression library included in many pieces of software such as Node.js, Bun, and Envoy.
The flaw could let an attacker run their own code on affected systems by sending or replying with malicious compressed data, without any need to create a user account. That is, an unpatched server that accepts compressed input could be taken over by a malicious client, and an unpatched client that accepts compressed input (such as “curl --compressed”) could be taken over by a malicious server.
This root cause issue was actually fixed in the Brotli source code in February, but it wasn’t known at the time that it gave rise to a security vulnerability and there was no more recent release of Brotli. Since Brotli is built into other software, any software which includes this library needs to take the update too, followed by the users of that software. We disclosed the vulnerability to Brotli and other downstream projects such as Node.js, which released or are currently releasing security updates, and Brotli cut new release versions and published a CVE.
More details on the patch and mitigations below:
patchtheplanet.org/brotli
Coming Soon
"AIs found zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents.
We used their discoveries to further strengthen our defenses. We resumed our frontier RL training."
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…
Dominic Kramer retweeted
Sam Altman says OpenAI's current best internal model is too powerful to release openly any time soon.
Before that could happen, the whole cyber defense setup of the internet would need to change and the world would need a new layer of protection against biological threats.
----
Full video on "Stanford Graduate School of Business" YouTube channel, (link in comment)
Simple solution. Use that internal model to solve that problem.
Sam Altman says OpenAI's current best internal model is too powerful to release openly any time soon.
Before that could happen, the whole cyber defense setup of the internet would need to change and the world would need a new layer of protection against biological threats.
----
Full video on "Stanford Graduate School of Business" YouTube channel, (link in comment)
Dominic Kramer retweeted
GPT-6 Astra Pro in ChatGPT got an upgrade with faster responses and improved use of plugins and tools.
And now, multiple agents can tackle different parts of a task in parallel. You can follow each agent’s progress and open completed tasks to review their work.
Now available to Pro users in the Chat tab.
plan your trips in Space!
I definitely thought it would sway much more favorably in the 2X direction across Pro 100/200/500 at the cost of bringing back 5h limits in order to smoothen the demand curve.
Dominic Kramer retweeted
Steering is a good release.
Dominic Kramer retweeted
My team and I have been hard at work on this update, send me your feedback!
Delegate work in Codex to your dot.
Your dot can now start work in Codex and follow up on existing threads, drawing on your ChatGPT conversations, Codex threads, and automations. It’s also better at deciding when to keep a thread going or start fresh.
Your dot can also review and edit Scheduled Tasks in ChatGPT Work, too. Ask what’s scheduled or update a recurring task as plans change.
My team and I have been hard at work on this update, send me your feedback!
Delegate work in Codex to your dot.
Your dot can now start work in Codex and follow up on existing threads, drawing on your ChatGPT conversations, Codex threads, and automations. It’s also better at deciding when to keep a thread going or start fresh.
Your dot can also review and edit Scheduled Tasks in ChatGPT Work, too. Ask what’s scheduled or update a recurring task as plans change.
Dominic Kramer retweeted
Ultrafast is rolling out today for GPT-6.1 Sol in the API, Codex, and ChatGPT Work.
Near-Astra intelligence at up to 8x faster speeds than Sol Standard, so you can build as fast as the ideas come.
Dominic Kramer retweeted
Day 4/
We have improved steering to be instant, leading to the model now reacting much faster to adjustments you make, allowing you to course-correct direction in realtime and not have the model waste effort.
Also releasing GPT-6.1 Sol ultrafast. The two work very well together.
Steering is a good release.
introducing Intelligent UI.
ChatGPT just got a lot more visual—introducing Intelligent UI.
Intelligent UI allows ChatGPT to answer quickly with fully interactive user interfaces that make everyday answers more visual.
It makes learning complex topics easier, and it quickly creates tools to solve a task right in the moment.
Oh, and also...GPT-6 is coming to ChatGPT. For everyone.
Save your meeting notes in Space! 📒
Your next meeting comes with a note-taker who can help with the follow-up.
The Meetings plugin takes notes for you and saves a personalized summary and next steps in ChatGPT Space, based on what ChatGPT knows about you and the work you’ve done together.
Keep notes private or share them with your team, then ask ChatGPT to update a project plan or draft a follow-up.
Available in beta for Pro and Business users in the ChatGPT desktop app on macOS. Enterprise is coming soon.
Download the desktop app, then search for “Meetings” in the plugin directory.
chatgpt.com/download/
Dominic Kramer retweeted
If you're building an AI personal assistant I _highly_ recommend using gpt-live-1 for a voice layer. Its architecture (duplex + delegation) is a generation ahead of other models available via API, and it is uniquely good at this application.
try it:
platform.openai.com/audio/li…
Dominic Kramer retweeted
We are entering a new era of discovery now:
openai.com/index/sharing-ai-…
