Conference on Applied Machine Learning for Information Security
Washington, DC
Joined September 2017
- Tweets579
- Following52
- Followers1.2K
- Likes183
That ends our technical sessions for CAMLIS 2026; thanks for joining us, whether online or in person! Wrapup starts in just a moment, and keep an eye on this account for information about our 10th(!) CAMLIS in 2027.
Camlis Org retweeted
As @CamlisOrg is wrapping up, reflecting how grateful I am for a tremendous community. The most fruitful plants blossom in rich soil and that's what gives @CamlisOrg its _sui generis_. I and our team at Abundant Security enjoy the benefits of that richness.
Thanks to @UdacityDave and @HZoete/@alxndrdavies for exemplary keynotes, to the authors of well-researched talks and posters, and all of our amazing volunteers for making CAMLIS what it is. I couldn't be more proud of what you've made in this community and event over the last decade.
Last but certainly not least, Mahmoud Abdelfattah wraps up our Guardrails and Defense track with "kNNGuard: Turning LLM Hidden Activations into a Training-Free Configurable Guardrail", bringing classical clustering and mechanistic interpretability together for better guardrails.
Andy Hsu presents our third talk in Guardrails and Defense, "What Counts as PII? Constitutional Targets for Broad, Efficient Identifier Detection" -- making detection of Personally Identifiable Information more repeatable, measurable, and robust.
Talk two of the Guardrails and Defense session is "Hardening ML Malware Classifiers Against Adversarial Evasion: Systematic Benchmarking and Covariate Shift Correction" with Mihai Nechita, making the next move in the malware detection cat-and-mouse game.
Our first talk of Guardrails and Defense starts now, with Tamas Voros presenting "NXDirection: Bringing the NX-bit to the Residual Stream", using mechanistic interpretability to prevent indirect prompt injection.
Our last (😭) session of CAMLIS 2026 starts in just a few minutes. Join us for four talks about Guardrails and Defense, followed by our traditional program conclusion, where much data shall be scienced.
Huge thank-you to La Dolce Cup coffee cart for keeping us all caffeinated with amazing espresso drinks all throughout CAMLIS!
If you're in person, get your orders in quick, they're leaving early today :)
Two hours until CAMLIS 2026 day 3 starts! Our final session on attacks and defenses starts in two hours!
That's it for Day 2! If you're here in-person, our poster session and happy hour starts in just a little while; otherwise we'll see everyone on livestream back at 9:30ET tomorrow for day 3, Guardrails and Defense!
Final talk in Security of Agents, and it's a good one: "Evaluating Agentic Learning Harness Capabilities Without Labels via the Scaling Hypothesis" tacking the question of how you evaluate a harness without brittle benchmarks or the circular reasoning of LLM-as-a-judge.
