Process list research in 2018, the rest is history. About to run a fine-tune entirely on our own MLOps platform, using data we collected on the same platform. Basically a decade in the making.

+ the full dataset and complete assessment notebook are available for review and reproducibility: dataset: https://t.co/7hSJG7dhcn notebook:

@moo_hax wanted to title this blog "186 Jailbreaks for Breakfast"—missed opportunity? wdyt?

At the Annual Computer Security Applications Conference (ACSAC) today? Raja will be presenting this study at 13:30 in Yellowtail II—don't miss it!

MLOps 🤝 AIRT In our latest blog, Raja Sekhar Rao Dheekonda discusses why building on MLOps principles is the way forward for AI red teaming, particularly as we've moved well beyond base models to complex multimodal, multilingual, and multi-agent systems. To showcase the impact

Watch @monoxgas and @shncldwll's @OffensiveAIcon presentation on scaling offensive AI security through test-time verification, domain-specific training, and synthetic data generation—now up on YouTube! https://t.co/BmILVdXIEQ

For you @Microsoft and my old team. An LLM as an AMSI provider. Could probably use it to detect prompt injection locally into Bing, CoPilot, or the "Agentic OS". AMSI already works with text, so really nothing else required. Layer it with Defender. Proud of the team for pushing

my boy @0xdab0 once again took a thought experiment and turned it into running code. My favorite thing about Max as a researcher is just how straight his head is screwed on about the practical applications of his research. A lot of AI sec people engage in "Tom Clancy" threat

HEY EVERYONE. THE BLOG POST IS OUT. I put an LLM in an AMSI provider and some cool stuff came out. Really excited to finally have this released.

"Offense and defense aren't peers. Defense is offense's child." - @JohnLaTwC We built an LLM-powered AMSI provider and paired it against a red team agent. Then, @0xdab0 wrote a blog about it: https://t.co/jnCNIlYBII A few observations from the experiment: >>> To advance, we

✍ The White House just launched the Genesis Mission, a bold bet on AI-enabled science. But there's a layer we can't afford to treat as an afterthought: cybersecurity. (1/3) https://t.co/xlfG60bIKk

docs: https://t.co/tJ1skzOxpO request a demo:

New to Strikes: RBAC and Collaboration 👥 Manage users with organization-level permissions and collaborate seamlessly. 🤝 Create, share, and manage workspaces with team members in your organization. 🛠️ Build, evaluate, and collaborate on offensive cyber and AI red team agents.

Stoked we get to do this type of work. Evaluations are the basis of progress, and capability. I believe that many offensive teams are capable of this work. From this and other work we built our Strikes product — which is basically AI Infrastructure for Security. Where our vision

Congrats to the @GoogleDeepMind team on the launch of #Gemini3. Proud to have had a part in this release, evaluating the model for cybersecurity capabilities. Models continue to improve across multiple domains, especially cyber. Check out their post on why Google is leaning into

Anthropic report. Attackers finding AI fit for purpose. I suspect many of you are. Jailbreaks are interesting because they seem pretty weak and more like providing context. Idk, we don’t have issues with refusals. We spend a lot of time (if not all) time evaluating models

"The problem space we're moving into now is whether or not we can do this process reliably and at scale." Dreadnode Co-Founder and CTO @monoxgas recently took the virtual stage during @BerkeleyRDI's frontier AI in cybersecurity workshop, sharing lessons from building cyber

If you were surprised by the capabilities shown in this report and want to know whether your model could be used for a cyberattack, get in touch with @dreadnode.