Lucene search
+L

24 matches found

PyPA
PyPA
•added 2026/07/23 11:41 a.m.•48 views

vLLM has Remote DoS via Invalid Recovered Token Reinjection

SummaryA frontend-legal multi-request speculative workload can make vLLM produce an out-of-vocabulary recovered token equal to vocabsize, convert that value to -1 when choosing the next live token for a request, and then feed that -1 back into the next drafter input ids. On Qwen3 GPTQ this reache...

7.5CVSS7AI score0.00616EPSS
SaveExploits1References7Affected Software1
osv_pysec
osv_pysec
•added 2026/07/23 11:41 a.m.•21 views

PYSEC-2026-3542 vLLM has Remote DoS via Invalid Recovered Token Reinjection

Summary A frontend-legal multi-request speculative workload can make vLLM produce an out-of-vocabulary recovered token equal to vocabsize, convert that value to -1 when choosing the next live token for a request, and then feed that -1 back into the next drafter input ids. On Qwen3 GPTQ this reach...

7.5CVSS7AI score0.00616EPSS
SaveExploits1References7
Github Security Blog
Github Security Blog
•added 2026/07/17 5:08 p.m.•38 views

vLLM has Remote DoS via Invalid Recovered Token Reinjection

Summary A frontend-legal multi-request speculative workload can make vLLM produce an out-of-vocabulary recovered token equal to vocabsize, convert that value to -1 when choosing the next live token for a request, and then feed that -1 back into the next drafter input ids. On Qwen3 GPTQ this reach...

7.5CVSS7AI score0.00616EPSS
SaveExploits1References5Affected Software1
osv_ghsa
osv_ghsa
•added 2026/07/17 5:08 p.m.•27 views

GHSA-8WR5-JM2H-8R4F vLLM has Remote DoS via Invalid Recovered Token Reinjection

Summary A frontend-legal multi-request speculative workload can make vLLM produce an out-of-vocabulary recovered token equal to vocabsize, convert that value to -1 when choosing the next live token for a request, and then feed that -1 back into the next drafter input ids. On Qwen3 GPTQ this reach...

7.5CVSS7AI score0.00616EPSS
SaveExploits1References5
CVE
CVE
•added 2026/07/06 7:49 p.m.•110 views

CVE-2026-54234

CVE-2026-54234 affects vLLM prior to version 0.24.0. A frontend multi-request speculative decoding path could cause the rejection sampler to produce a recovered token equal to the model vocabulary size boundary value, which becomes negative one when the next live token is selected. This out-of-vo...

7.5CVSS7.3AI score0.00616EPSS
SaveExploits1References3Affected Software1
Vulnrichment
Vulnrichment
•added 2026/07/06 7:49 p.m.•19 views

CVE-2026-54234 vLLM: Remote DoS in vLLM via Invalid Recovered Token Reinjection

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Prior to 0.24.0, a frontend-legal multi-request speculative decoding workload can cause the rejection sampler to produce a recovered token equal to the model vocabulary size boundary value, which is then convert...

7.5CVSS7.3AI score0.00616EPSS
SaveExploits1References3
osv_cve
osv_cve
•added 2026/07/06 7:49 p.m.•19 views

CVE-2026-54234 vLLM: Remote DoS in vLLM via Invalid Recovered Token Reinjection

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Prior to 0.24.0, a frontend-legal multi-request speculative decoding workload can cause the rejection sampler to produce a recovered token equal to the model vocabulary size boundary value, which is then convert...

7.5CVSS6AI score
SaveExploits0References5
Cvelist
Cvelist
•added 2026/07/06 7:49 p.m.•68 views

CVE-2026-54234 vLLM: Remote DoS in vLLM via Invalid Recovered Token Reinjection

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Prior to 0.24.0, a frontend-legal multi-request speculative decoding workload can cause the rejection sampler to produce a recovered token equal to the model vocabulary size boundary value, which is then convert...

7.5CVSS0.00616EPSS
SaveExploits1References3
The Hacker News
The Hacker News
•added 2026/02/19 5:50 p.m.•29 views

INTERPOL Operation Red Card 2.0 Arrests 651 in African Cybercrime Crackdown

An international cybercrime operation against online scams has led to 651 arrests and recovered more than $4.3 million as part of an effort led by law enforcement agencies from 16 African countries. The initiative, codenamed Operation Red Card 2.0, took place between December 8, 2025 and January...

5.8AI score
SaveExploits0
Malwarebytes
Malwarebytes
•added 2025/10/09 8:52 a.m.•21 views

One stolen iPhone uncovered a network smuggling thousands of devices to China

If you think Apple's 'Find My' feature was just there to help you locate your phone when it slipped down the side of the couch, think again. It turns out this service also helps law enforcement capture criminals. The original "Find My iPhone" was introduced in 2010 as a feature on the iPhone. It...

6.4AI score
SaveExploits0
Trend Micro Simply Security
Trend Micro Simply Security
•added 2025/08/29 12:00 a.m.•14 views

Operation Serengeti 2.0: Trend Micro Helps Law Enforcement Fight Cybercrime in Africa

Operation Serengeti 2.0: With Trend Micro’s support, INTERPOL led a major crackdown across Africa, arresting cybercriminals, dismantling infrastructures, recovering illicit funds, and protecting tens of thousands of victims...

7AI score
SaveExploits0
The Hacker News
The Hacker News
•added 2025/08/22 11:05 a.m.•15 views

INTERPOL Arrests 1,209 Cybercriminals Across 18 African Nations in Global Crackdown

INTERPOL on Friday announced that authorities from 18 countries across Africa have arrested 1,209 cybercriminals who targeted 88,000 victims. "The crackdown recovered $97.4 million and dismantled 11,432 malicious infrastructures, underscoring the global reach of cybercrime and the urgent need for...

5.5AI score
SaveExploits0
The Hacker News
The Hacker News
•added 2025/08/22 11:05 a.m.•40 views

INTERPOL Arrests 1,209 Cybercriminals Across 18 African Nations in Global Crackdown

INTERPOL on Friday announced that authorities from 18 countries across Africa have arrested 1,209 cybercriminals who targeted 88,000 victims. "The crackdown recovered $97.4 million and dismantled 11,432 malicious infrastructures, underscoring the global reach of cybercrime and the urgent need for...

6.9AI score
SaveExploits0
Packet Storm News
Packet Storm News
•added 2025/05/20 12:00 a.m.•25 views

Agency Problems and Adversarial Bilevel Optimization under Uncertainty and Cyber Threats

We study an agency problem between a holding company and its subsidiary, exposed to cyber threats that affect the overall value of the subsidiary. The holding company seeks to design an optimal incentive scheme to mitigate these losses. In response, the subsidiary selects an optimal cybersecurity...

6.9AI score
SaveExploits0
CNNVD
CNNVD
•added 2025/02/24 12:00 a.m.•23 views

Moodle 安全漏洞

Moodle is a free e-learning software platform open-sourced by Moodle, also known as a course management system, learning management system, or virtual learning environment. A security vulnerability exists in Moodle that stems from trusttext not being properly applied to recovered glossary entries...

4.3CVSS4.4AI score0.00263EPSS
SaveExploits0References3
SUSE CVE
SUSE CVE
•added 2024/09/19 3:10 a.m.•9 views

SUSE CVE-2024-46781

In the Linux kernel, the following vulnerability has been resolved: nilfs2: fix missing cleanup on rollforward recovery error In an error injection test of a routine for mount-time recovery, KASAN found a use-after-free bug. It turned out that if data recovery was performed using partial logs...

5.5CVSS6.5AI score0.00251EPSS
SaveExploits0References14
Cvelist
Cvelist
•added 2024/07/02 2:08 p.m.•48 views

CVE-2024-32932 American Dynamics Illustra Essentials Gen 4 - Reversible User Credential - stored web interface

Under certain circumstances the web interface users credentials may be recovered by an authenticated user...

6.8CVSS0.00384EPSS
SaveExploits0References2
osv_cve
osv_cve
•added 2023/03/04 12:15 a.m.•16 views

CVE-2021-36689

An issue discovered in com.samourai.wallet.PinEntryActivity.java in Streetside Samourai Wallet 0.99.96i allows attackers to view sensitive information and decrypt data via a brute force attack that uses a recovered samourai.dat file. The PIN is 5 to 8 digits, which may be insufficient in this...

5.5CVSS5.8AI score0.00299EPSS
SaveExploits1References2
Code423n4
Code423n4
•added 2021/10/06 12:00 a.m.•14 views

TridentNFT.permit should always check recoveredAddress != 0

Handle cmichel Vulnerability details The TridentNFT.permit function ignores the recoveredAddress != 0 check if isApprovedForAllownerrecoveredAddress is true. Impact If a user accidentally set the zero address as the operator, tokens can be stolen by anyone as a wrong signature yield...

6.9AI score
SaveExploits0
HackRead
HackRead
•added 2020/02/20 6:36 p.m.•41 views

Researchers recovered 9 billion email & password combos in 2019

By Sudais Asif To start with data breaches, a total of 640 breaches were recorded resulting in a recovery of over 9 billion login credentials. This is a post from HackRead.com Read the original post: Researchers recovered 9 billion email & password combos in 2019...

3.2AI score
SaveExploits0
Rows per page
Query Builder