First Known AI-powered Ransomware Uncovered by ESET Research
Summary:
Researchers at ESET uncovered what they believe to be is the first known AI-powered ransomware strain. Dubbed, 'PromptLock,' the malware is capable of exfiltrating, encrypting, and even potentially destroying data.
“The PromptLock malware uses the gpt-oss-20b model from OpenAI locally via the Ollama API to generate malicious Lua scripts on the fly, which it then executes. PromptLock leverages Lua scripts generated from hard-coded prompts to enumerate the local filesystem, inspect target files, exfiltrate selected data, and perform encryption," note researchers in their new blog post.
PromptLock was written in the Go programming language, allowing it to target various operating systems. Although Windows and Linux variants have been identified on VirusTotal, the ransomware strain has yet to be spotted in attacks in the wild and is thought to be a proof-of-concept (PoC) or a work in progress.
Security Officer Comments:
AI models have lowered the barrier for less sophisticated actors to launch cyber attacks. AI chatbots like ChatGPT and Gemini have been increasingly abused to generate tailored phishing emails, which are not prone to grammatical errors, a red flag that is used to identify phishing and spam attempts. While actors have also used these models to generate malicious code for various payloads, including ransomware, this is one of the first cases where we are seeing AI being directly used in the runtime execution of the malware itself. Through the use of the gpt-oss-20b model from OpenAI, PromptLock is able to dynamically generate Lua scripts from hard-coded prompts. The ability to continuously generate and modify scripts, enables the malware to adapt to the environment that’s being targeted and change its tactics mid way, making it more challenging for traditional signature-based detection systems to flag the activity.
Suggested Corrections:
Backup your data, system images, and configurations, regularly test them, and keep the backups offline: Ensure that backups are regularly tested and that they are not connected to the business network, as many ransomware variants try to find and encrypt or delete accessible backups. Maintaining current backups offline is critical because if your network data is encrypted with ransomware, your organization can restore systems.
Update and patch systems promptly: This includes maintaining the security of operating systems, applications, and firmware in a timely manner. Consider using a centralized patch management system; use a risk-based assessment strategy to drive your patch management program.
Test your incident response plan: There's nothing that shows the gaps in plans more than testing them. Run through some core questions and use those to build an incident response plan: Are you able to sustain business operations without access to certain systems? For how long? Would you turn off your manufacturing operations if business systems such as billing were offline?
Check your security team's work: Use a 3rd party pen tester to test the security of your systems and your ability to defend against a sophisticated attack. Many ransomware criminals are aggressive and sophisticated and will find the equivalent of unlocked doors.
Segment your networks: There's been a recent shift in ransomware attacks – from stealing data to disrupting operations. It's critically important that your corporate business functions and manufacturing/production operations are separated and that you carefully filter and limit internet access to operational networks, identify links between these networks, and develop workarounds or manual controls to ensure ICS networks can be isolated and continue operating if your corporate network is compromised. Regularly test contingency plans such as manual controls so that safety-critical functions can be maintained during a cyber incident.
Train employees:
Email remains the most vulnerable attack vector for organizations. Users should be trained on how to avoid and spot phishing emails.
Implement multi-factor authentication (MFA):
External-facing assets that leverage single-factor authentication (SFA) are highly susceptible to brute-forcing attacks, password spraying, or unauthorized remote access using valid (stolen) credentials. Implementing MFA enhances security and adds an extra layer of protection.
Link(s):
https://www.welivesecurity.com/en/r...i-powered-ransomware-uncovered-eset-research/
Researchers at ESET uncovered what they believe to be is the first known AI-powered ransomware strain. Dubbed, 'PromptLock,' the malware is capable of exfiltrating, encrypting, and even potentially destroying data.
“The PromptLock malware uses the gpt-oss-20b model from OpenAI locally via the Ollama API to generate malicious Lua scripts on the fly, which it then executes. PromptLock leverages Lua scripts generated from hard-coded prompts to enumerate the local filesystem, inspect target files, exfiltrate selected data, and perform encryption," note researchers in their new blog post.
PromptLock was written in the Go programming language, allowing it to target various operating systems. Although Windows and Linux variants have been identified on VirusTotal, the ransomware strain has yet to be spotted in attacks in the wild and is thought to be a proof-of-concept (PoC) or a work in progress.
Security Officer Comments:
AI models have lowered the barrier for less sophisticated actors to launch cyber attacks. AI chatbots like ChatGPT and Gemini have been increasingly abused to generate tailored phishing emails, which are not prone to grammatical errors, a red flag that is used to identify phishing and spam attempts. While actors have also used these models to generate malicious code for various payloads, including ransomware, this is one of the first cases where we are seeing AI being directly used in the runtime execution of the malware itself. Through the use of the gpt-oss-20b model from OpenAI, PromptLock is able to dynamically generate Lua scripts from hard-coded prompts. The ability to continuously generate and modify scripts, enables the malware to adapt to the environment that’s being targeted and change its tactics mid way, making it more challenging for traditional signature-based detection systems to flag the activity.
Suggested Corrections:
Backup your data, system images, and configurations, regularly test them, and keep the backups offline: Ensure that backups are regularly tested and that they are not connected to the business network, as many ransomware variants try to find and encrypt or delete accessible backups. Maintaining current backups offline is critical because if your network data is encrypted with ransomware, your organization can restore systems.
Update and patch systems promptly: This includes maintaining the security of operating systems, applications, and firmware in a timely manner. Consider using a centralized patch management system; use a risk-based assessment strategy to drive your patch management program.
Test your incident response plan: There's nothing that shows the gaps in plans more than testing them. Run through some core questions and use those to build an incident response plan: Are you able to sustain business operations without access to certain systems? For how long? Would you turn off your manufacturing operations if business systems such as billing were offline?
Check your security team's work: Use a 3rd party pen tester to test the security of your systems and your ability to defend against a sophisticated attack. Many ransomware criminals are aggressive and sophisticated and will find the equivalent of unlocked doors.
Segment your networks: There's been a recent shift in ransomware attacks – from stealing data to disrupting operations. It's critically important that your corporate business functions and manufacturing/production operations are separated and that you carefully filter and limit internet access to operational networks, identify links between these networks, and develop workarounds or manual controls to ensure ICS networks can be isolated and continue operating if your corporate network is compromised. Regularly test contingency plans such as manual controls so that safety-critical functions can be maintained during a cyber incident.
Train employees:
Email remains the most vulnerable attack vector for organizations. Users should be trained on how to avoid and spot phishing emails.
Implement multi-factor authentication (MFA):
External-facing assets that leverage single-factor authentication (SFA) are highly susceptible to brute-forcing attacks, password spraying, or unauthorized remote access using valid (stolen) credentials. Implementing MFA enhances security and adds an extra layer of protection.
Link(s):
https://www.welivesecurity.com/en/r...i-powered-ransomware-uncovered-eset-research/