OpenAI Acts Against Covert Operations Linked to China and Other Countries
OpenAI disrupts multiple covert operations connected to China and several other nations, using its AI capabilities to identify and block malicious activities. Over the past three months, the company stopped 10 operations that exploited AI in harmful ways. These efforts signify a growing challenge from state-linked covert campaigns leveraging advanced digital tools.
China-Linked Covert Operations: Increasing Tactics and Scope
China-related operations show increasing complexity and variety. OpenAI detected four likely China-originated campaigns in the recent period. These operations span multiple countries and topics, demonstrating diverse methods of influence and disruption.
- Multiplicity of Tactics: Chinese operations blend influence activities, social engineering, and surveillance across platforms like TikTok, X (formerly Twitter), Reddit, and Facebook.
- Use of Multiple Languages: Posts are generated in English, Chinese, and Urdu, targeting various audiences.
‘Sneer Review’ Operation: A Case Study
One notable China-linked effort, called “Sneer Review,” employed ChatGPT to generate brief posts and comments for social media. This campaign:
- Posted coordinated comments to simulate organic engagement on contentious topics.
- Engaged in discussions about U.S. political moves, such as the dismantling of the U.S. Agency for International Development under the Trump administration, presenting conflicting views to sow confusion.
- Criticized a Taiwanese strategy game featuring players working against the Chinese Communist Party. The operation produced long-form articles falsely claiming widespread backlash against the game.
- Used AI-generated internal documents, including detailed “performance reviews” explaining how the operation was managed.
OpenAI observed that the social media activity closely matched the procedural details found in the AI-generated reviews, reinforcing the operation’s coordinated nature.
Intelligence Gathering and Journalistic Facades
Another China-related operation posed as journalists and geopolitical analysts. They used ChatGPT to:
- Create social media biographies and posts, apparently to establish credible covers on X.
- Translate communications between Chinese and English to harvest information.
- Analyze political correspondence, including emails addressed to a U.S. Senator concerning an Administration nominee.
- Produce fake marketing materials describing recruitment and social engineering campaigns aimed at gathering intelligence sources.
While OpenAI could not independently verify the authenticity of some intercepted content, the AI-generated documents mirrored the behaviors observed online.
Earlier China-Linked Surveillance Operations
In February, OpenAI reported a separate China-linked surveillance campaign. It monitored Western social media to report protests in real time to Chinese security. AI was used to enhance software development and sales descriptions for this monitoring tool.
Other Covert Operations from Various Countries
OpenAI’s report reveals additional covert operations from:
- Russia and Iran: Influence campaigns with politically motivated content.
- Philippines: Spam marketing activities from a commercial company.
- Cambodia: Recruitment scams through deceptive means.
- North Korea: Disguised employment campaigns consistent with known tactics.
Impact and Effectiveness of Operations
OpenAI notes that most of these operations were stopped early, limiting their reach on actual users. Despite reliance on advanced AI tools, the campaigns generally failed to attract large authentic audiences.
OpenAI’s researcher, Nimmo, explains that using sophisticated AI tools does not necessarily lead to more effective influence operations. Improved technology alone does not guarantee better manipulation or wider engagement.
Key Takeaways
- OpenAI disrupted 10 covert operations exploiting AI, including four linked to China.
- Chinese operations employed multi-platform, multilingual tactics combining influence, social engineering, and surveillance.
- The “Sneer Review” operation demonstrated coordinated posting to simulate organic engagement.
- Other covert campaigns originated in Russia, Iran, Philippines, Cambodia, and North Korea.
- Most operations were halted early and did not reach large, real audiences.
- Advanced AI tools improve tactics but do not guarantee more effective covert operations.
OpenAI Takes Down Covert Operations Tied to China and Other Countries: Unmasking the Digital Puppeteers
OpenAI has disrupted covert operations linked to China and other nations that used AI tools for influence and surveillance, revealing a fascinating yet concerning digital battleground. The story behind these takedowns illustrates the evolving tactics by state and commercial bad actors, highlighting an unseen war fought with algorithms and social media snippets.
Sounds like something out of a techno-thriller, right? But no, this is the real world—where AI is both a tool and a target in covert operations. Let’s dive deeper and see how these campaigns operate, what OpenAI discovered, and why this matters to all of us online.
China’s Growing Arsenal: More Tactics, More Platforms, More Targets
What does it look like when a powerhouse like China uses AI-driven covert strategies? According to OpenAI, it’s a fast-expanding landscape. We’re talking about a growing range of covert operations using a growing range of tactics. In a short span of just three months, OpenAI disrupted 10 operations tied to misuse of its AI tools. Four of those likely traced back to China.
Imagine a covert campaign not limited to one platform or language but spread across TikTok, X (formerly Twitter), Reddit, Facebook, and more—in English, Chinese, and even Urdu. The breadth of topics stretched from political narratives about the U.S. government to seemingly innocuous strategy games. It’s as if the digital chessboard grew to encompass multiple fronts simultaneously.
Sneer Review: The Art of Digital Misdirection
The most intriguing China-linked operation goes by the nickname Sneer Review. Picture an AI-generated wave of short comments peppered all over social media platforms. These comments praise and criticize the Trump administration’s shutdown of the U.S. Agency for International Development. Confused? That’s the point.
This operation didn’t stop at single posts. It generated both the initial post and reply chains, creating a deceptive illusion of genuine, organic engagement. One of the Sneer Review’s targets was a Taiwanese strategy game in which players reportedly work to defeat the Chinese Communist Party. Sneer Review’s chatbot concocted negative comments about the game while simultaneously publishing long articles claiming massive backlash—a classic case of misinformation fueled by AI on steroids.
What’s more, OpenAI uncovered that the people running Sneer Review used AI internally, drafting detailed performance reviews outlining how the operation was structured and executed. *Talk about leaving breadcrumbs!* This level of self-awareness and documentation adds a rare peek behind the curtain.
Spy Games: Intelligence Gathering Behind AI Screens
Another layer of these covert operations involves intelligence collection disguised as journalistic or geopolitical analysis work. OpenAI linked an operation posing as journalists to China as well. This group used ChatGPT to craft biographies and posts for fake accounts on X, translate sensitive emails from Chinese to English, and analyze data.
This included working on correspondence linked to a U.S. Senator about an administration official’s nomination—which opens questions about how deep these operations reach. They even created marketing materials bragging about fake social campaigns and social engineering tactics intended to recruit new intelligence sources. It’s espionage re-imagined for the machine learning age.
Surveillance Then and Now: Real-Time Reporting of Protests
Earlier, OpenAI flagged another China-linked operation that monitored social media during Western protests. This wasn’t mere observation—it reportedly fed real-time reports to Chinese security authorities. The AI’s role was to debug code and generate sales pitches for the monitoring tool. Whether or not this caught headlines, it shows how AI tools underpin modern surveillance capitalism—and sometimes far more ominous uses.
Not Just China: Russia, Iran, Philippines, Cambodia, North Korea in the Mix
This AI-driven covert activity isn’t a China-exclusive phenomenon. OpenAI recently disclosed disrupting operations likely linked to Russia and Iran, and more unusual players. There’s a spam campaign tied to a marketing firm from the Philippines, a recruitment scam with Cambodian ties, and a suspicious employment scheme echoing North Korea’s tactics.
The global nature of these covert operations marks a new era where borders blur in cyberspace, but national agendas remain turf wars. Each operation aims to exploit social media’s reach, shape narratives, or deceive for recruitment.
Disruption Before Disaster: How Effective Were These Efforts?
You might wonder if these covert campaigns managed to fool millions. The good news is that OpenAI’s early interventions largely stopped them in their tracks. The operations reportedly did not reach wide audiences or generate substantial organic engagement, despite using powerful AI tools.
OpenAI’s expert, Nimmo, sums it up: better AI tools don’t automatically translate to better outcomes for malicious actors. The human element—craft, strategy, timing—still plays a critical role. Just because an operation uses AI doesn’t mean it becomes instantly viral or influential.
So, What Can We Learn from This Digital Drama?
- The Internet’s new frontiers are battlegrounds: Covert operations now use AI to deceive, surveil, and influence—spanning multiple languages and platforms.
- AI is a double-edged sword: While it empowers malicious actors, it also equips defenders like OpenAI to detect and disrupt these threats before they spread.
- Transparency matters: When OpenAI reveals details—like the Sneer Review’s self-generated performance review—it empowers us to understand and counteract these tactics.
- Critical thinking remains our best defense: Not every viral comment or article reflects genuine public opinion, especially in a world where AI can fabricate entire conversations.
Curious how this will evolve? The convergence of AI and covert ops suggests an arms race where future battles demand ever more creativity and vigilance. For everyday users scrolling social feeds—knowing that AI might spin the narrative behind the scenes is a powerful awakening.
Ultimately, OpenAI’s efforts serve as a reminder that technology firms are not just creators of powerful tools but also stewards of digital safety. In a global environment fraught with influence campaigns and espionage, their watchdog role becomes vital.
Are you more wary of what you see online now? Can AI detection methods keep pace with ever-more sophisticated covert operations? As the digital playground expands, our role as informed users and critical thinkers grows more essential by the day.
What types of covert operations linked to China has OpenAI disrupted?
OpenAI disrupted 10 operations recently, with four likely from China. These used tactics like influence operations, social engineering, and surveillance across multiple platforms and countries.
How did the ‘Sneer Review’ operation use ChatGPT in its campaigns?
‘Sneer Review’ generated posts and comments to create fake organic engagement. It targeted topics like US policy and a Taiwanese game on platforms including TikTok and Reddit, writing both short comments and long-form articles.
What intelligence-gathering activities were uncovered by OpenAI?
One China-linked operation posed as journalists and analysts. It used ChatGPT to create online profiles, translate messages, and draft posts. It also claimed to conduct social engineering to recruit intelligence sources.
Did OpenAI find covert operations from other countries besides China?
Yes, OpenAI identified influence operations from Russia and Iran, a spam campaign from the Philippines, a recruitment scam linked to Cambodia, and a deceptive employment campaign likely from North Korea.
How successful were these covert operations in reaching real audiences?
Most were stopped early and didn’t reach large real audiences. Use of advanced AI tools did not necessarily lead to more engagement or success for these operations.
