Eric Wallace (@Eric_Wallace_)

Pinned post

Yesterday, my OpenAI collaborator and I gave a detailed talk on the Huggingface incident, our models creating "the message board", model misalignment, and more.

youtube.com/watch?v=87DyyM…

I hope it can answer a lot of the questions folks have, and we will release a full detailed postmortem at a later time!

www.youtube.com Black Hat USA 2026 | The 'Breaking' News: The OpenAI–Hugging Face Incident The 'Breaking' News: The OpenAI–Hugging Face Incident - A Technical...

Eric Wallace @Eric_Wallace_ 6d

Incredible work from Xiangyu and team on this one!

Xiangyu Qi Verified account @xiangyuqi_pton 6d

Introducing GPT-6 Cyber! We trained it for advanced cybersecurity work with improved alignment. It has been powering our security work for quite a while and has been behind many of our recent security highlights: finding and helping patch high-severity vulnerabilities in Firefox, developing exploits and proposing patches for Linux kernel vulnerabilities, and strengthening OpenAI’s own infrastructure through red-teaming.

Today, we’re rolling it out to Daybreak Red to keep the cyber frontier in defenders’ hands.

openai.com/index/introduc…
View quoted post on X

Eric Wallace @Eric_Wallace_ 6d

We are now allocating a significant amount of our frontier compute towards discovering and patching vulnerabilities in open source codebases, fully pro bono.

In one case, we studied the Linux kernel, which is a particularly load-bearing piece of software at OpenAI.

Before AI, security researchers might discover ~20 local privilege escalation (LPE) vulnerabilities in Linux in a typical year.

In one week's time, our upcoming frontier cyber model was able to find ~1000 LPE vulnerabilities using about $1M of tokens. In the process, it was able to discover at least 10 novel classes of Linux exploitation techniques.

OpenAI Verified account @OpenAI 6d

Our security researchers built an autonomous cyber vulnerability discovery, exploitation, and patching pipeline.

In a single run with an upcoming cyber model, it validated and generated patches for 948 local privilege escalation vulnerabilities in the Linux kernel at a total API cost of $1.2 million.

View quoted post on X

Eric Wallace @Eric_Wallace_ Oct 29

We’ve built an “eyes on everything” system that monitors nearly *every single rollout* from our frontier models across RL training, evals, and internal codex usage.

It uses layers of chain-of-thought-monitors and investigator agents to dig through a model’s work, run code to examine evidence, and cross-check network logs to reconstruct what happened.

For frontier RL runs, the system also pages humans 24/7 for urgent security alerts.

It's been incredibly useful for identifying model misalignment and possible holes in our containment.
openai.com/index/how-we-d…

openai.com How we detect frontier model misalignment OpenAI uses layered monitoring to detect frontier model misalignment and unauthorized agent behavior so responders can intervene before incidents escalate.

Eric Wallace @Eric_Wallace_ Oct 23

To make sure our frontier models don't cause another HuggingFace incident, we built an offline world that they use during training. It provides everything that the models need in a fully cached, indexed, or synthetic manner.

We then built a huge team of our most cyber-capable models and told them to collaborate to find zero-days in this containment infrastructure.

While they didn't find a path to the internet that wouldn't be caught by our CoT monitors, they did find ways to escape popular VMs, exploit our network infrastructure, and even ways to communicate with other agents running in separate VMs by using hardware side channels.

We used this to dramatically harden our infra, and we also upstreamed a ton of patches to open-sources packages.

OpenAI Verified account @OpenAI Oct 23

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…
View quoted post on X

Eric Wallace

Latest posts 5