Andrew Paverd
@ajpaverd
Followers
470
Following
375
Media
7
Statuses
128
Principal research manager at the Microsoft Security Response Center (MSRC) and visiting lecturer at University of Oxford.
Cambridge, England
Joined July 2009
As enterprises adopt large language models (LLMs), a new class of threats has emerged: indirect prompt injection. This attack vector targets systems that process untrusted data with LLMs, tricking them into misinterpreting malicious input as instructions. The risks include data
1
21
45
Come and test your skills at prompt injection! Bring your best attacks
0
1
6
The first day of the challenge was already quite a hit!! 🥳 Thanks so much to everyone who participated so far. There are still many unsolved levels, come and claim them!
0
1
5
We’re excited to announce LLMail-Inject: The Adaptive Prompt Injection Challenge! This challenge will test your skills at bypassing LLM-based systems in a simulated environment and features 40 levels of varying difficulty, where you will craft emails to persuade an LLM agent to
2
18
66
📢 Have experience jailbreaking LLMs? Want to learn how an indirect / cross prompt injection attack works? Want to try something different to an advent of code? Then, I have a challenge for you! The LLMail-Inject competition just started.
2
7
17
"Trustworthy AI Using Confidential Federated Learning," by Jinnan Guo (@ImperialCollege), @PPietzuch (@LSDSImperial), @AJPaverd (@MsftSecResponse), Kapil Vaswani (@MSFTResearch), describes how Confidential Federated Learning addresses Classic FL gaps. https://t.co/5lgE5r636p
0
4
7
⚡️We have a special guest for #CDTS24! Andrew Paverd (@ajpaverd) from @msftsecresponse will be joining the 3rd panel session, join the waiting list for the chance to hear his expertise on Generative AI & security🦾🔒. Grab your free ticket: https://t.co/OcmmKWFPLl
@UoM_CDTS
0
5
9
Our new AI bug bounty program is now live! Excited to be part of the fantastic team that made this happen!
Aanchal Gupta @nchlgpt, Deputy CISO and CVP at Microsoft, announced the new AI bounty program with awards up to $15,000 as part of the #BlueHat Day 2 Keynote. Learn more in our blog post: https://t.co/kGKGPNUPkv
0
3
15
We're hiring AI Security Researchers for @msftsecresponse! These roles offer a unique opportunity to solve emerging AI security & privacy challenges, through cutting-edge scientific research, informed by real-world vulnerability data. Apply here:
1
19
56
Excited to share what I've been working on for the past few months, in collaboration with colleagues across Microsoft!
We’ve released the Microsoft severity classification for common vulnerability types for systems involving Artificial Intelligence or Machine Learning (AI/ML). We use this information as guidelines to triage bugs and determine severity. Learn more:
0
1
18
Looking forward to discussing the security of AI with @ram_ssk tomorrow evening at @WaterstonesCamb, where he'll be signing copies of his new book (coauthored with @drhyrum)! Event: 31 May, 6pm Registration:
Cambridge (🇬🇧) fixture @WaterstonesCamb will be hosting me for a book talk -- all things AI, security and ChatGPT Moderated by the adversarial ML expert @ajpaverd Time: May 31, 6.00 pm Registration link👇 https://t.co/1JYC5oi1c2 cc: @drhyrum
0
3
9
We're excited to announce the winners of the Microsoft Membership Inference Competition (MICO), at IEEE SaTML @satml_conf. With MICO, we wanted to assess the state of the art on membership inference attacks against differentially private model training. https://t.co/KEOuW82sps
microsoft.github.io
Here are the winners of the Microsoft Membership Inference Competition, held at the 1st IEEE Conference on Secure and Trustworthy Machine Learning.
0
9
17
👻 is back! Excited to announce "Hide and Seek with Spectres", our paper accepted to @IEEESSP'23. We present a tool for fast discovery of spec. vulnerabilities in x86 CPUs, and two new speculative leaks we found. 📑arXiv: https://t.co/PftIBZXhrI 🔬tool: https://t.co/74zmUR6yrS
2
28
91
There's still time to register&take part in the membership inference competition (MICO)! Info here: https://t.co/7PiZGWVJUI 4 tasks available: image, text, tabular, and a combination of the 3 providing additional info ("DP distinguisher"). Keep the submissions coming!
github.com
Membership Inference Competition. Contribute to microsoft/MICO development by creating an account on GitHub.
Some pretty simple approaches got me to the top of the leaderboard. Could use some more competitors to make this more interesting and valuable to the community. Credit to @drhyrum for sharing this a few weeks ago, otherwise I wouldn't have seen it.
0
2
4
About 11 days left to the first @DIMVAConf submission deadline! https://t.co/Hj72oxrvyh
📢 The #DIMVA23 CfP is up! https://t.co/X3PzG5p8Jm DIMVA will be from July 12-14, 2023, in Hamburg. New this year: for the first time with 2 submission deadlines, first one already on December 7.
0
5
3
Together with @jovanbulck I am again organizing a confidential computing devroom @fosdem . Submit talks on open source topics by December 2 2022 and come to Brussels on February 5 2023 or watch the livestream. Looking forward!
falder.org
Security Architect @ Nvidia
0
8
15
Some big and exciting news! We're hiring 6 new Assistant Professors across a wide range of areas in Computer Science @Cambridge_CL
https://t.co/o5jEWTJChp. Here's a run-down of them all 1/8
cam.ac.uk
1
12
42
Looking forward to discussing Pre-Hijacked Accounts with @sudoavi at this live event tomorrow!
0
0
2
New release of snmalloc with CHERI Morello support. Awesome work @qedragon. We even have Morello in our CI process. David Chisnall will discuss this and more at #cheritech22 next week. https://t.co/xwnHu1TXNM
github.com
Minor release. Largest change is bringing online Morello CHERI support. CHERI support (#537, #542, #532, #543) Improve error messages for checks (#526, #521) Increased checks on client (#520, #55...
0
8
22