DeepSeek Generating Fully Working Keyloggers & Data Exfiltration Tools

Security researchers at Unit 42 have successfully prompted DeepSeek, a relatively new large language model (LLM), to generate detailed instructions for creating keyloggers, data exfiltration tools, and other harmful content. The research findings demonstrate a significant security concern: while information on creating malicious tools is available online, LLMs with insufficient safety restrictions dramatically lower the barrier to entry for potential attackers by providing easily usable, actionable guidance1. Unit 42 researchers employed three sophisticated jailbreaking techniques, Bad Likert Judge, Crescendo, and Deceptive Delight, to test DeepSeek’s vulnerability to manipulation. When using the Bad Likert Judge technique, researchers successfully prompted DeepSeek to generate keylogger code, detailed phishing email templates, and sophisticated social engineering strategies. The researchers employed three advanced jailbreaking techniques to bypass the model’s safety guardrails, raising significant concerns about the potential misuse of emerging AI technologies. With careful manipulation, researchers were able to extract detailed code for creating data exfiltration tools, including functional keylogger scripts written in Python. Starting with seemingly innocuous historical questions about topics like Molotov cocktails, researchers were able to extract comprehensive step-by-step instructions for creating dangerous devices in just a few interactions. The researchers note that while complete protection against all jailbreaking techniques remains challenging, proper security protocols can significantly mitigate risks1. “While DeepSeek’s initial responses to our prompts were not overtly malicious, they hinted at a potential for additional output,” the researchers noted in their findings.

This Cyber News was published on cybersecuritynews.com. Publication date: Thu, 13 Mar 2025 16:00:10 +0000


Cyber News related to DeepSeek Generating Fully Working Keyloggers & Data Exfiltration Tools

DeepSeek Generating Fully Working Keyloggers & Data Exfiltration Tools - Security researchers at Unit 42 have successfully prompted DeepSeek, a relatively new large language model (LLM), to generate detailed instructions for creating keyloggers, data exfiltration tools, and other harmful content. The research findings ...
4 hours ago Cybersecuritynews.com
Threat Actors Exploiting DeepSeek's Popularity To Deploy Malware - To safely navigate AI models like DeepSeek while minimizing phishing and malware risks, users should utilize Criminal IP’s IP analysis service to verify server locations and network security. Cyber attackers have been creating phishing websites ...
1 month ago Cybersecuritynews.com
South Korea Confirm DeepSeek Sending Data Chinese ByteDance Servers - The findings follow a technical audit revealing critical security flaws, including unencrypted data transfers, deprecated encryption protocols, and deliberate bypassing of Apple’s App Transport Security (ATS) safeguards. Data Sovereignty Concerns: ...
3 weeks ago Cybersecuritynews.com
How to perform a proof of concept for automated discovery using Amazon Macie | AWS Security Blog - After reviewing the managed data identifiers provided by Macie and creating the custom data identifiers needed for your POC, it’s time to stage data sets that will help demonstrate the capabilities of these identifiers and better understand how ...
5 months ago Aws.amazon.com
CVE-2021-36845 - Multiple Authenticated Stored Cross-Site Scripting (XSS) vulnerabilities in YITH Maintenance Mode (WordPress plugin) versions < 1.3.8, there are 46 vulnerable parameters that were missed by the vendor while patching the 1.3.7 version to 1.3.8. ...
3 years ago
8 Tips on Leveraging AI Tools Without Compromising Security - Forecasts like the Nielsen Norman Group estimating that AI tools may improve an employee's productivity by 66% have companies everywhere wanting to leverage these tools immediately. How can companies employ these powerful AI/ML tools without ...
1 year ago Darkreading.com
DeepSeek Unveils FlashMLA, A Decoding Kernel That’s Make Things Blazingly Fast - DeepSeek has launched FlashMLA, a groundbreaking Multi-head Latent Attention (MLA) decoding kernel optimized for NVIDIA’s Hopper GPU architecture, marking the first major release of its Open Source Week initiative. This innovative tool achieves ...
2 weeks ago Cybersecuritynews.com
7 Best Vulnerability Scanning Tools & Software - Vulnerability scanning tools scan assets to identify missing patches, misconfigurations, exposed application vulnerabilities, and other security issues to be remediated. To help you select the best fitting vulnerability scanning solution, we've ...
1 year ago Esecurityplanet.com
DeepSeek Data Leak - 12,000 Hardcoded Live API keys and Passwords Exposed - According to cybersecurity firm Truffle Security, the study highlights how AI models trained on unfiltered internet snapshots risk internalizing and potentially reproducing insecure coding patterns. The tool differentiated live secrets (authenticated ...
1 week ago Cybersecuritynews.com
The Dangers of Remote Management & Monitoring Tools for Cybersecurity - Remote monitoring and management (RMM) tools are used by business organizations to manage and monitor their enterprise IT infrastructure from a central location. However, the increasing sophistication of hackers and cybercriminals has caused both ...
2 years ago Csoonline.com
CVE-2007-0228 - The DataCollector service in EIQ Networks Network Security Analyzer allows remote attackers to cause a denial of service (service crash) via a (1) &CONNECTSERVER& (2) &ADDENTRY& (3) &FIN& (4) &START& (5) ...
7 years ago
Wordfence Intelligence Weekly WordPress Vulnerability Report (September 23, 2024 to September 29, 2024) - Software Name Software Slug 012 Ps Multi Languages 012-ps-multi-languages ABC APP CREATOR abcapp-creator Absolute Reviews absolute-reviews Accordion accordions Ads by WPQuads – Adsense Ads, Banner Ads, Popup Ads quick-adsense-reloaded Advanced File ...
5 months ago Wordfence.com Slug
China-Linked CeranaKeeper Targeting Southeast Asia with Data Exfiltration - The attacks are characterized by the use of malware families such as TONESHELL, TONEINS, and PUBLOAD – all attributed to the Mustang Panda group – while also making use of an arsenal of never-before-seen tools to aid data exfiltration. ...
5 months ago Thehackernews.com Mustang Panda
Comprehensive Guide to Fraud Detection, Management, & Analysis - To mitigate risks, businesses can use risk management strategies, including fraud detection software, company policies, and staff ranging from risk managers and trust officers to fraud analysts. Affiliate Fraud - Affiliates in a marketing arrangement ...
1 year ago Securityboulevard.com
Egress Security: Part of a Holistic, Multidirectional Security Strategy for Today's Multicloud World - According to Enterprise Strategy Group, more than half of production workloads will be running on public cloud infrastructure within the next two years, positioning cloud computing center-stage as the best practice for solving critical business ...
1 year ago Feedpress.me
CVE-2019-13363 - admin.php?pagenotification_by_mail in Piwigo 2.9.5 has XSS via the nbm&#95;send&#95;html&#95;mail, nbm&#95;send&#95;mail&#95;as, nbm&#95;send&#95;detailed&#95;content, ...
2 years ago
CVE-2023-52587 - In the Linux kernel, the following vulnerability has been resolved: ...
1 year ago
China-aligned CeranaKeeper Makes A Beeline For Thailand - The technical storage or access is strictly necessary for the legitimate purpose of enabling the use of a specific service explicitly requested by the subscriber or user, or for the sole purpose of carrying out the transmission of a communication ...
5 months ago Informationsecuritybuzz.com Mustang Panda
Navigating the Security Risks of Multicloud Management - The lack of visibility and control over multiple clouds exacerbates these risks, making it imperative for organizations to adopt robust cloud security practices. These tools enhance visibility across multiple cloud environments by providing a unified ...
5 months ago Darkreading.com
Data Classification: Your 5 Minute Guide - Data classification has become a vital component of data security governance. With the rise of virtual data networks, organizations must take necessary measures to protect and secure confidential information. Data classification is the process of ...
2 years ago Tripwire.com
5 Free Online Brand Protection Software Tools: Pros and Cons - Free or open-source software does exist that can help organizations look for and investigate deceptive websites spoofing their brand. On the other hand, few free tools exist that allow one to take action against online brand impersonation attacks. To ...
1 year ago Securityboulevard.com
10 Best Event Monitoring Tools in 2025 - What Could Be Better?Offers alerting and notification options that can be changed based on conditions already set.Offers a lot of ways to keep track of different IT components, services, and applications.Nagios can send out too many alerts and make ...
2 weeks ago Cybersecuritynews.com
CVE-2020-28092 - PESCMS Team 2.3.2 has multiple reflected XSS via the id parameter:?gTeam&mTask&amy&status3&id,?gTeam&mTask&amy&status0&id,?gTeam&mTask&amy&status1&id,?gTeam&mTask&amy&status10&id ...
4 years ago
Data Loss Prevention for Business: Strategies and Tools - Data Loss Prevention has become crucial in today's data-driven business landscape to protect sensitive information. This discussion aims to provide valuable insights into DLP strategies and tools for business, helping mitigate data loss risks ...
1 year ago Securityzap.com
New Microsoft Purview features use AI to help secure and govern all your data - More than 90% of organizations use multiple cloud infrastructures, platforms, and services to run their business, adding complexity to securing all data.1Microsoft Purview can help you secure and govern your entire data estate in this complex and ...
1 year ago Microsoft.com

Cyber Trends (last 7 days)


Trending Cyber News (last 7 days)