Eraser: Jailbreaking Defense in Large Language Models via Unlearning Harmful Knowledge

Fuente: arXiv

Gespeichert in:

Bibliographische Detailangaben
Hauptverfasser:	Lu, Weikai, Zeng, Ziqian, Wang, Jianwei, Lu, Zhengdong, Chen, Zelin, Zhuang, Huiping, Chen, Cen
Format:	Preprint
Veröffentlicht:	2024
Schlagworte:	Computation and Language
Online-Zugang:	Acceder al recurso
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Ähnliche Einträge

SEA: Low-Resource Safety Alignment for Multimodal Large Language Models via Synthetic Embeddings
von: Lu, Weikai, et al.
Veröffentlicht: (2025)

Decompose, Plan in Parallel, and Merge: A Novel Paradigm for Large Language Models based Planning with Multiple Constraints
von: Lu, Zhengdong, et al.
Veröffentlicht: (2025)

RCP-Merging: Merging Long Chain-of-Thought Models with Domain-Specific Models by Considering Reasoning Capability as Prior
von: Yang, Junyao, et al.
Veröffentlicht: (2025)

PrivacyRestore: Privacy-Preserving Inference in Large Language Models via Privacy Removal and Restoration
von: Zeng, Ziqian, et al.
Veröffentlicht: (2024)

ConsistentEE: A Consistent and Hardness-Guided Early Exiting Method for Accelerating Language Models Inference
von: Zeng, Ziqian, et al.
Veröffentlicht: (2023)

GenderAlign: An Alignment Dataset for Mitigating Gender Bias in Large Language Models
von: Zhang, Tao, et al.
Veröffentlicht: (2024)

Dissecting Fine-Tuning Unlearning in Large Language Models
von: Hong, Yihuai, et al.
Veröffentlicht: (2024)

Chimera: A Lossless Decoding Method for Accelerating Large Language Models Inference by Fusing all Tokens
von: Zeng, Ziqian, et al.
Veröffentlicht: (2024)

E$^2$AT: Multimodal Jailbreak Defense via Dynamic Joint Optimization for Multimodal Large Language Models
von: Lu, Liming, et al.
Veröffentlicht: (2025)

SynAdapt: Learning Adaptive Reasoning in Large Language Models via Synthetic Continuous Chain-of-Thought
von: Wang, Jianwei, et al.
Veröffentlicht: (2025)

Jigsaw Puzzles: Splitting Harmful Questions to Jailbreak Large Language Models
von: Yang, Hao, et al.
Veröffentlicht: (2024)

Retrieval-Augmented Defense: Adaptive and Controllable Jailbreak Prevention for Large Language Models
von: Yang, Guangyu, et al.
Veröffentlicht: (2025)

Provable Defense Framework for LLM Jailbreaks via Noise-Augumented Alignment
von: Cheng, Zehua, et al.
Veröffentlicht: (2026)

Learning with Silver Standard Data for Zero-shot Relation Extraction
von: Wang, Tianyin, et al.
Veröffentlicht: (2022)

Unified Defense for Large Language Models against Jailbreak and Fine-Tuning Attacks in Education
von: Yi, Xin, et al.
Veröffentlicht: (2025)

JailbreakZoo: Survey, Landscapes, and Horizons in Jailbreaking Large Language and Vision-Language Models
von: Jin, Haibo, et al.
Veröffentlicht: (2024)

Knowledge-to-Jailbreak: Investigating Knowledge-driven Jailbreaking Attacks for Large Language Models
von: Tu, Shangqing, et al.
Veröffentlicht: (2024)

RewardDS: Privacy-Preserving Fine-Tuning for Large Language Models via Reward Driven Data Synthesis
von: Wang, Jianwei, et al.
Veröffentlicht: (2025)

ARGUS: Defending Against Multimodal Indirect Prompt Injection via Steering Instruction-Following Behavior
von: Lu, Weikai, et al.
Veröffentlicht: (2025)

On the use of Silver Standard Data for Zero-shot Classification Tasks in Information Extraction
von: Wang, Jianwei, et al.
Veröffentlicht: (2024)

CCJA: Context-Coherent Jailbreak Attack for Aligned Large Language Models
von: Zhou, Guanghao, et al.
Veröffentlicht: (2025)

Resolving Editing-Unlearning Conflicts: A Knowledge Codebook Framework for Large Language Model Updating
von: Zhang, Binchi, et al.
Veröffentlicht: (2025)

THRD: A Training-Free Multi-Turn Defense Framework for Jailbreak Attacks on Large Language Models
von: Ma, Zhiqing, et al.
Veröffentlicht: (2026)

Model Editing Harms General Abilities of Large Language Models: Regularization to the Rescue
von: Gu, Jia-Chen, et al.
Veröffentlicht: (2024)

To Forget or Not? Towards Practical Knowledge Unlearning for Large Language Models
von: Tian, Bozhong, et al.
Veröffentlicht: (2024)

InfoFlood: Jailbreaking Large Language Models with Information Overload
von: Yadav, Advait, et al.
Veröffentlicht: (2025)

JailbreakLens: Visual Analysis of Jailbreak Attacks Against Large Language Models
von: Feng, Yingchaojie, et al.
Veröffentlicht: (2024)

Jailbreaking Commercial Black-Box LLMs with Explicitly Harmful Prompts
von: Zhang, Chiyu, et al.
Veröffentlicht: (2025)

LLM-Align: Utilizing Large Language Models for Entity Alignment in Knowledge Graphs
von: Chen, Xuan, et al.
Veröffentlicht: (2024)

SDD: Self-Degraded Defense against Malicious Fine-tuning
von: Chen, Zixuan, et al.
Veröffentlicht: (2025)

Audio Jailbreaks in Large Audio-Language Models: Taxonomy, Attack-Defense Analysis, and Cost-Aware Evaluation
von: Feng, Bo-Han, et al.
Veröffentlicht: (2026)

Foot In The Door: Understanding Large Language Model Jailbreaking via Cognitive Psychology
von: Wang, Zhenhua, et al.
Veröffentlicht: (2024)

ID-Eraser: Proactive Defense Against Face Swapping via Identity Perturbation
von: Luo, Junyan, et al.
Veröffentlicht: (2026)

Jailbreaking Multimodal Large Language Models via Shuffle Inconsistency
von: Zhao, Shiji, et al.
Veröffentlicht: (2025)

Learning and Unlearning of Fabricated Knowledge in Language Models
von: Sun, Chen, et al.
Veröffentlicht: (2024)

EvoDefense: Co-Evolving Black-Box Defense with Large Language Models
von: Li, Yu, et al.
Veröffentlicht: (2026)

AutoDefense: Multi-Agent LLM Defense against Jailbreak Attacks
von: Zeng, Yifan, et al.
Veröffentlicht: (2024)

RWKU: Benchmarking Real-World Knowledge Unlearning for Large Language Models
von: Jin, Zhuoran, et al.
Veröffentlicht: (2024)

ZeroUnlearn: Few-Shot Knowledge Unlearning in Large Language Models
von: Lin, Yujie, et al.
Veröffentlicht: (2026)

Concept Unlearning in Large Language Models via Self-Constructed Knowledge Triplets
von: Yamashita, Tomoya, et al.
Veröffentlicht: (2025)

if (filemtime($filename) + $maxlifetime < time()) { unlink($filename); $count++; } } return $count; } /** * A function that is called internally when session data is to be saved. * * @param string $sessId The current session ID * @param string $data The session data to write * * @return bool */ protected function saveSession($sessId, $data): bool { $sessFile = $this->path . '/sess_' . $sessId; if ($handle = fopen($sessFile, 'w')) { $return = false; // Lock the file for exclusive access to avoid issues with multiple // processes writing session simultaneously: if (flock($handle, LOCK_EX)) { $return = fwrite($handle, $data); // Make sure that there's no trailing data by truncating the file to // the correct length: ftruncate($handle, strlen($data)); fflush($handle); flock($handle, LOCK_UN); } fclose($handle); if ($return !== false) { return true; } } // If we got this far, something went wrong with the file output... // It is tempting to throw an exception here, but this code is called // outside of the context of exception handling, so all we can do is // echo a message.

// Anecdotal testing Today and Yesterday seems to indicate destroy() // is called by the garbage collector and everything is good. // Something to keep in mind though. return 0; } /** * Write function that is called when session data is to be saved. * * @param string $sessId The current session ID * @param string $data The session data to write * * @return bool */ public function write($sessId, $data): bool { if ($this->writesDisabled) { return true; } return $this->saveSession($sessId, $data); } /** * A function that is called internally when session data is to be saved. * * @param string $sessId The current session ID * @param string $data The session data to write * * @return bool */ abstract protected function saveSession($sessId, $data): bool; }

* * @return void */ public function writeClose() { // The assumption is that we're using PHP's ext/session. // session_write_close() will actually overwrite $_SESSION with an // empty array on completion -- which leads to a mismatch between what // is in the storage object and $_SESSION. To get around this, we // temporarily reset $_SESSION to an array, and then re-link it to // the storage object. // // Additionally, while you _can_ write to $_SESSION following a // session_write_close() operation, no changes made to it will be // flushed to the session handler. As such, we now mark the storage // object isImmutable. $storage = $this->getStorage(); if (! $storage->isImmutable()) { $_SESSION = $storage->toArray(true); session_write_close(); $storage->fromArray($_SESSION); $storage->markImmutable(); } } /** * Attempt to set the session name * * If the session has already been started, or if the name provided fails * validation, an exception will be raised. * * @param string $name * @return SessionManager * @throws Exception\InvalidArgumentException */ public function setName($name) { if ($this->sessionExists()) { throw new Exception\InvalidArgumentException( 'Cannot set session name after a session has already started'

Environment & details:

GET Data empty

POST Data empty

Files empty

Key	Value
VUFIND_SESSION	"8e9c674bfa6c224f50f0861b8975c5e8"
ui	"standard"
language	"de"

Session

Key	Value
__Laminas	array:2 [`"_REQUEST_ACCESS_TIME" => 1790882347.8796 "_VALID" => array:1 [ "Laminas\Session\Validator\Id" => "8e9c674bfa6c224f50f0861b8975c5e8" ]`]
SessionState	Laminas\Stdlib\ArrayObject {#938}
FlashMessenger	Laminas\Stdlib\ArrayObject {#1433}
SessionHelper	Laminas\Stdlib\ArrayObject {#1759}

Server/Request Data

Key	Value
REDIRECT_VUFIND_LOCAL_DIR	"/usr/local/vufind/local"
REDIRECT_HTTPS	"on"
REDIRECT_SSL_TLS_SNI	"info.umar.mx"
REDIRECT_STATUS	"200"
VUFIND_LOCAL_DIR	"/usr/local/vufind/local"
HTTPS	"on"
SSL_TLS_SNI	"info.umar.mx"
HTTP_ACCEPT	"/"
HTTP_USER_AGENT	"Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; ClaudeBot/1.0; +claudebot@anthropic.com)"
HTTP_COOKIE	"VUFIND_SESSION=8e9c674bfa6c224f50f0861b8975c5e8; ui=standard; language=de"
HTTP_ACCEPT_ENCODING	"gzip, br, zstd, deflate"
HTTP_HOST	"info.umar.mx"
PATH	"/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/snap/bin"
SERVER_SIGNATURE	"<address>Apache/2.4.66 (Ubuntu) Server at info.umar.mx Port 443</address>\n"
SERVER_SOFTWARE	"Apache/2.4.66 (Ubuntu)"
SERVER_NAME	"info.umar.mx"
SERVER_ADDR	"200.23.222.19"
SERVER_PORT	"443"
REMOTE_ADDR	"216.73.217.142"
DOCUMENT_ROOT	"/usr/local/vufind/public"
REQUEST_SCHEME	"https"
CONTEXT_PREFIX	""
CONTEXT_DOCUMENT_ROOT	"/usr/local/vufind/public"
SERVER_ADMIN	"admin@localhost"
SCRIPT_FILENAME	"/usr/local/vufind/public/index.php"
REMOTE_PORT	"62650"
REDIRECT_URL	""" /Record/\n arxiv_https___arxiv_org_abs_2404_05880/Similar """
GATEWAY_INTERFACE	"CGI/1.1"
SERVER_PROTOCOL	"HTTP/1.1"
REQUEST_METHOD	"GET"
QUERY_STRING	""
REQUEST_URI	"/Record/%0Aarxiv_https___arxiv_org_abs_2404_05880/Similar"
SCRIPT_NAME	"/index.php"
PHP_SELF	"/index.php"
REQUEST_TIME_FLOAT	1790882347.8753
REQUEST_TIME	1790882347

Environment Variables empty

Registered Handlers

0. Whoops\Handler\PrettyPageHandler