@ditrizna Profile picture

Dmitrijs Trizna

@ditrizna

Independent Cyber-Security and AI Researcher | ex-Microsoft. I have two Master's Degrees, almost Ph.D., OSCP, GREM, GDAT, etc.

Similar User
Jonny Johnson photo

@jsecurity101

Outflank photo

@OutflankNL

Matt Hand photo

@matterpreter

Enno Rey photo

@Enno_Insinuator

Connected photo

@LucyIsOpal

Aaron Jornet photo

@RexorVc0

bohops photo

@bohops

Steven photo

@0xthirteen

Cody Thomas photo

@its_a_feature_

Chris Thompson photo

@retBandit

Antonio Cocomazzi photo

@splinter_code

Jason Lang photo

@curi0usJack

Cn33liz photo

@Cneelis

monoxgas photo

@monoxgas

Scott Sutherland photo

@_nullbind

Pinned

@MSFTBlueHat organized awesome event for the first time coming in India — was happy to be part of it! Recording of the talk is coming ✌️

Dmitrijs Trizna (@ditrizna), Senior Security Researcher at Microsoft, is presenting his #BlueHatIndia talk: “The Impact of Backdoor Poisoning Vulnerabilities on AI-Based Threat Detectors.” In his talk, Dmitrijs discussed AI-based defenses: threat model: living-off-the-land, data…

Tweet Image 1
Tweet Image 2
Tweet Image 3


This Notebook should be of interest to any AI Red Teamer, since has all the details about SOTA model that is evaded only with “human interpretable” adversarial perturbations!

A stand-alone replication Colab for **Ensemble Everything Everywhere** is out! 🎉 ⚔️You train & attack a SOTA-like multi-resolution self-ensemble from our paper /Ensemble everything everywhere: Multi-scale aggregation for adversarial robustness/ 🔗 Links in the first reply!



This is awesome role and team! If you are in US, consider this position working with @Cyb3rWard0g!

📣 As part of the Autonomous Defense & Protection Team (ADAPT), my team is growing and taking shape with recent hires, and I’m excited to announce a new opportunity to join us! 🔥Apply here ➡️ jobs.careers.microsoft.com/global/en/job/… 🚀 I’m hiring a Principal Applied Scientist to support our…



Hah, this is gold!

Periodic Table of Windows Events #ThreatHunting #DFIR #Windows

Tweet Image 1


That’s why I say to people: “LLM’s will not find you 0-days.” LLMs indeed can be helpful to identify known weaknesses present in the training dataset, or as N-shot examples in RAG, but as @niklas2484 et al. show, LLMs will fail to generalize even on those examples.

LLMs recently achieved impressive results on vulnerability detection datasets -- but how good are they really? Paper: arxiv.org/abs/2306.17193 Code: github.com/niklasrisse/Li… w/ @mboehme_ (1/13)

Tweet Image 1


Dmitrijs Trizna Reposted

If you're reading this tweet while in @BlackHatEvents, join me and @ditrizna in our journey through ML malware detection and adversarial pentesting! 😈 August 3/4 and 5/6! See you there 😎 blackhat.com/us-24/training… #blackhat #malware #detection #ai #machinelearning


Loading...

Something went wrong.


Something went wrong.