OpenAI Shuts Down Accounts Used to Generate Phishing Emails & Malware

While Artificial Intelligence holds immense potential for good, its power can also attract those with malicious intent. 

State-affiliated actors, with their advanced resources and expertise, pose a unique threat, leveraging AI for cyberattacks that can disrupt infrastructure, steal data, and even harm individuals.

“We terminated accounts associated with state-affiliated threat actors. Our findings show our models offer only limited, incremental capabilities for malicious cybersecurity tasks.”

OpenAI teamed up with Microsoft Threat Intelligence to disrupt five state-affiliated groups attempting to misuse their AI services for malicious activities.

Document
Live Account Takeover Attack Simulation

How do Hackers Bypass 2FA?

Live attack simulation Webinar demonstrates various ways in which account takeover can happen and practices to protect your websites and APIs against ATO attacks .

State-affiliated groups

Two groups linked to China, known as Charcoal Typhoon and Salmon Typhoon,

The Iranian threat actor “Crimson Sandstorm,” North Korea’s “Emerald Sleet,” and Russia-affiliated group “Forest Blizzard.”

Charcoal Typhoon: Researched companies and cybersecurity tools, likely for phishing campaigns.

Salmon Typhoon: Translated technical papers, gathered intelligence on agencies and threats, and researched hiding malicious processes.

Crimson Sandstorm: Developed scripts for app and web development, crafted potential spear-phishing content, and explored malware detection evasion techniques.

Emerald Sleet: Identified security experts, researched vulnerabilities, assisted with basic scripting, and drafted potential phishing content.

Forest Blizzard: Conducted open-source research on satellite communication and radar technology while also using AI for scripting tasks.

OpenAI’s latest security assessments, conducted with experts, show that while malicious actors attempt to misuse AI like GPT-4, its capabilities for harmful cyberattacks remain relatively basic compared to readily available non-AI tools.

OpenAI strategy

Proactive Defense: actively monitor and disrupt state-backed actors misusing platforms with dedicated teams and technology.

Industry Collaboration: work with partners to share information and develop collective responses against malicious AI use.

Continuously Learning: analyze real-world misuse to improve safety measures and stay ahead of evolving threats.

Public Transparency: share insights about malicious AI activity and actions to promote awareness and preparedness.

Stay updated on Cybersecurity news, Whitepapers, and Infographics. Follow us on LinkedIn & Twitter.

Gurubaran

Gurubaran is a co-founder of Cyber Security News and GBHackers On Security. He has 10+ years of experience as a Security Consultant, Editor, and Analyst in cybersecurity, technology, and communications.

Recent Posts

New VIPKeyLogger Via Weaponized Office Documenrs Steals Login Credentials

The VIPKeyLogger infostealer, exhibiting similarities to the Snake Keylogger, is actively circulating through phishing campaigns. …

4 hours ago

INTERPOL Urges to End ‘Pig Butchering’ & Replaces With “Romance Baiting”

INTERPOL has called for the term "romance baiting" to replace "pig butchering," a phrase widely…

5 hours ago

New I2PRAT Malware Using encrypted peer-to-peer communication to Evade Detections

Cybersecurity experts are sounding the alarm over a new strain of malware dubbed "I2PRAT," which…

6 hours ago

Earth Koshchei Employs RDP Relay, Rogue RDP server in Server Attacks

 A new cyber campaign by the advanced persistent threat (APT) group Earth Koshchei has brought…

7 hours ago

Careto – A legendary Threat Group Targets Windows By Deploy Microphone Recorder And Steal Files

Recent research has linked a series of cyberattacks to The Mask group, as one notable…

7 hours ago

RiseLoader Attack Windows By Employed A VMProtect To Drop Multiple Malware Families

RiseLoader, a new malware family discovered in October 2024, leverages a custom TCP-based binary protocol…

7 hours ago