Pierluigi Paganini
June 09, 2025
OpenAI banned ChatGPT accounts that were used by Russian-speaking threat actors and two Chinese nation-state actors. The blocked accounts were used to assist malware development, social media automation, and research about U.S. satellite communications technologies, among other malicious activities.
The AI firm banned a set of accounts used by a Russian-speaking group as part of a cyber operation tracked as ScopeCreep. The group used the tool to develop Windows malware and establish a stealthy C2 infrastructure.
The actor focused on stealthy tactics, distributing the malware via a trojanized gaming tool and evading detection using obfuscation, DLL sideloading, and Telegram alerts. The attackers used multiple temporary accounts, each for a single code tweak, showing strong operational security. OpenAI’s detection and collaboration with partners led to account bans and takedown of the malicious code repository before widespread impact.
“At this stage, the impact of ScopeCreep may have been mitigated by quick reporting and close collaboration with industry partners who were able to take down the malicious repository. We banned the OpenAI accounts used by this adversary.” reads the report published by OpenAI. “We assess this threat actor utilized our models in an attempt to speed up their malware development operations. Paradoxically, this also provided an opportunity for us to identify and disrupt the threat quickly and in what looked like its early stages.”
The company also banned accounts linked to a likely Russia-based influence operation, tracked as Helgoland Bite, targeting Germany’s 2025 election. The actors used ChatGPT to generate German-language content critical of the U.S. and NATO, shared via Telegram and X. Posts supported the AfD party and were echoed by the Russian-linked Pravda network. Despite broad distribution and 27K followers on X, authentic engagement was low, placing the operation at the high end of Category 2 for influence impact.
The company also banned accounts tied to a likely China-based influence operation dubbed “Sneer Review.” Threat actor used the accounts to mass-produce social media posts in Chinese, English, and Urdu, often political in nature, and internal performance documents. The posts, shared on TikTok, X, Reddit, and Facebook, included fake engagement to simulate popularity and targeted topics like Taiwan, USAID, and activist Mahrang Baloch. The company pointed out that despite their tactics, the operation appeared to be in its early stages with limited authentic reach. OpenAI detected and disrupted the activity before significant impact occurred.
The company banned a small network of accounts linked to “VAGue Focus,” a likely China-origin operation blending social engineering and influence tactics. The actors used OpenAI platform to create fake personas, craft emails, and translate outreach messages posing as media or consultancy groups targeting the U.S. and Europe. Some content offered payment for classified info. Though posing as professionals, the operation had little authentic engagement. Activity suggests low sophistication and limited impact.
OpenAI banned accounts linked to Chinese threat actors VIXEN PANDA (APT15) and KEYHOLE PANDA (APT5), who used ChatGPT in various stages of cyber operations. These actors used the model for tasks like script modification, system configuration, penetration testing automation, and Android-based social media manipulation. They also researched U.S. defense and telecom infrastructure. Though their activity was broad, the AI didn’t give them capabilities beyond what public tools already offer.
Another interesting operation blocked by the AI firm is the Operation “Uncle Spam”, which was a China-linked influence campaign that used ChatGPT to create polarizing U.S. political content, targeting both sides of debates like tariffs. The actors generated AI-written posts and images, crafted fake veteran personas, and sought tools to scrape personal data from platforms like X and Bluesky. Despite broad efforts across platforms, the operation saw minimal authentic engagement and was classified as low-impact (Category 2 on the IO scale).
“These operations originated in many parts of the world, acted in many different ways, and focused on many different targets. A significant number appeared to originate in China: Four of the 10 cases in this report, spanning social engineering, covert influence operations and cyber threats, likely had a Chinese origin.” concludes the report. “But we’ve disrupted abuses from many other countries too: this report includes case studies of a likely task scam from Cambodia, comment spamming apparently from the Philippines, covert influence attempts potentially linked with Russia and Iran, and deceptive employment schemes.”
Follow me on Twitter: @securityaffairs and Facebook and Mastodon
(SecurityAffairs – hacking, ChatGPT accounts)
Security / November 03, 2025
