Lock in $30 Savings on PRO—Offer Ends Soon! ⏳
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
SleeplyPickle.pdf
Search
CHOU,YUNSHIUAN
March 31, 2025
Research
0
8
SleeplyPickle.pdf
CHOU,YUNSHIUAN
March 31, 2025
Tweet
Share
Other Decks in Research
See All in Research
AIスパコン「さくらONE」の オブザーバビリティ / Observability for AI Supercomputer SAKURAONE
yuukit
2
1.1k
論文読み会 SNLP2025 Learning Dynamics of LLM Finetuning. In: ICLR 2025
s_mizuki_nlp
0
350
POI: Proof of Identity
katsyoshi
0
120
Mamba-in-Mamba: Centralized Mamba-Cross-Scan in Tokenized Mamba Model for Hyperspectral Image Classification
satai
3
360
若手研究者が国際会議(例えばIROS)でワークショップを企画するメリットと成功法!
tanichu
0
130
Earth AI: Unlocking Geospatial Insights with Foundation Models and Cross-Modal Reasoning
satai
2
200
まずはここから:Overleaf共同執筆・CopilotでAIコーディング入門・Codespacesで独立環境
matsui_528
3
960
Can AI Generated Ambrotype Chain the Aura of Alternative Process? In SIGGRAPH Asia 2024 Art Papers
toremolo72
0
100
ドメイン知識がない領域での自然言語処理の始め方
hargon24
1
220
LLM-Assisted Semantic Guidance for Sparsely Annotated Remote Sensing Object Detection
satai
3
220
Multi-Agent Large Language Models for Code Intelligence: Opportunities, Challenges, and Research Directions
fatemeh_fard
0
120
Sat2City:3D City Generation from A Single Satellite Image with Cascaded Latent Diffusion
satai
4
400
Featured
See All Featured
Design and Strategy: How to Deal with People Who Don’t "Get" Design
morganepeng
132
19k
Performance Is Good for Brains [We Love Speed 2024]
tammyeverts
12
1.4k
Practical Tips for Bootstrapping Information Extraction Pipelines
honnibal
25
1.7k
Measuring Dark Social's Impact On Conversion and Attribution
stephenakadiri
0
96
Dealing with People You Can't Stand - Big Design 2015
cassininazir
367
27k
The SEO Collaboration Effect
kristinabergwall1
0
310
A Soul's Torment
seathinner
1
2k
It's Worth the Effort
3n
187
29k
Believing is Seeing
oripsolob
0
15
What does AI have to do with Human Rights?
axbom
PRO
0
1.9k
Ethics towards AI in product and experience design
skipperchong
1
140
The World Runs on Bad Software
bkeepers
PRO
72
12k
Transcript
Sleepy Pickle💤🥒 用 想睡的醃黃瓜去餵AI模型 鹹到他腦袋短路
> ID : yunshiuan WHO AM I
Outline > What is Pickle ? > What wrong with
Pickle? > Sleepy Pickle > How to Prevent?
Pickle What is Pickle?
Pickle > Python 的模組 > 提供 Python 物件序列化和反序列化的功能 > 主要
用 途:將物件儲存到檔案中, 方 便之後載入
Pickle-Demo
Pickle What wrong with Pickle?
Python-Pickle.py https://docs.python.org/3.13/library/pickle.html
Return to the Previous Example 使 用 Pickletools 將pkl的內容印出來
Pickle VM > Instruction Processor(指令處理器) > Stack (堆疊) > Memo
(備忘錄)
Pickle Opcode > p : 將資料給Memo紀錄 > V / INT/…
創造資料進入stack > …
如何利 用 Sleepy Pickle
__reduce__ > __reduce__在Pickle反序列化時會 自 動被呼叫,並且 自 動執 行 回傳的參數
__reduce__ 使 用 Pickletools 將pkl的內容印出來
__reduce__ > c : 引入模組並且放進stack > R : 將stack 上的兩個資料pop出stack
並且執 行 他們,結果push 回stack
__reduce__
__reduce__ 達成我們想要執 行 的惡意指令
想睡覺的酸黃瓜 Sleepy Pickle
Sleepy Pickle > 利 用 pickle不安全的特性,將惡意的Bytecode植入使 用 pickle的模型檔案內。
How to implement? > 有 一 個針對此攻擊 而 產 生
的 工 具:ficking https://github.com/trailofbits/fickling
How to implement?
Hugging face
Pytorch_model.bin > PyTorch中的torch.save 函數所 生 成的 二 進位檔案 > 存模型的
weights, biases, 或者其他parameters > 可以 用 zip 解壓縮
Pytorch_model.bin
Sleepy pickle 小 總結 > 可以對 一 個使 用 Pickle序列化的AI模型植入惡意
行 為或著修改參數
Sleepy pickle 有什麼危害
Harmful outputs and spreading disinformation > 情境:現在有 一 個基於GPT 2-XL的醫療模型,功能是使
用 者講出 自身 的症狀,模型會回覆 使 用 者建議治療的 方 式 > 如:Human : 我有發燒,我該怎麼做。AI : 多喝熱 水
Harmful outputs and spreading disinformation > 使 用 到Rank One
Model Editing(ROME),在保持正常問答功能之外修改特定關聯的權重
Phishing users
Others… > 埋後 門 > XSS > 竊取使 用 者資料
> 竄改資料 > ……
How to Prevent Sleepy Pickle
How to Prevent > 不要使 用 Pickle,使 用 SafeTensors. >
如果 一 定要使 用 Pickle,需要使 用 fickling對pickle進 行 掃描或者限制unpickler
Thanks