RAM Speedwagon
Welcome to Memetic Warfare.
This week we’ll check out Open AI’s latest report. This one is a bit shorter than usual and evokes a bit of self-interest, covering China-linked influence operations targeting “AI debates in the US”:
The summary is pretty concise:
The first operation they named “data center bandwagon”, when the very obvious RAM Bandwagon was right there. I personally would have gone with “make data center haters your waiters” had I not already used that reference in a previous blogpost. Happy to consult on future operation names guys! If you’re not going to do a relevant pop culture reference, at least make it a chengyu or something.
The operation is somewhat straightforward: China-origin accounts used ChatGPT to create content to denigrate data centers. These operators also made the unusual choice to upload their strategic planning documents to ChatGPT, presumably for better prompt results?
They used ChatGPT to create obviously AI-generated content describing how electricity bills are higher now, apparently due to data centers. They sought to make this content popular amongst Americans by using the almost household name hashtag of #capacityauction. They also used real coverage from local newspapers for authenticity.
Some other posts were even less inspired:
Other hashtags they tried to launch include #datacentersuccess and #datacenters, the last of which probably at least has some discourse on it.
Another cluster tried to target Chinese dissidents, something that many Chinese operations and Spamouflage in particular do:
Some accounts portrayed themselves as Chinese Americans or Chinese citizens living in America:
It’s interesting to see how the same network used ChatGPT for for their operational security and platform detection bypass methods.
Because these operators just don’t care, we gain insight into how they view bypassing Facebook’s safeguards:
Open AI named the next operation “Tech and Tariffs”, and I can already think of some better names.
This time, it’s personal:
This operation got up to a few different types of activity, including criticizing US tech and tariff policy, trying to design social listening systems and - gasp - editing work reports.
Open AI says that they used terminology similar to how China’s public security apparatus works - thus insinuating ties to the MPS or perhaps SHanghai police, which has also dabbled in similar stuff.
The most interesting part is the targeting of Open AI itself. The operation accused ChatGPT of suffering a data breach, exposing personal information of ChatGPT users:
They criticize Trump and praise China’s activity:
They’re not above dipping their toes into antisemitic content:
Funnily enough, they do the same thing that Spamouflage has done before: make the president look cool.
They also tried to get ChatGPT to propose a plan for an “AI system” for social listening. This is like the fourth time this has happened so far.
Other audiences were targeted as well:
One of the accounts also got into the cyber attribution space:
That user posted a timeline of alleged Western operations and tooling:
While it’s low-res, it seems to be translated from Chinese and presumably is an AI-generated timeline from past CVERC or other reporting. My initial thought of the most similar content was the China Cybersecurity Industry Alliance’s “report” on “mobile cyberattacks”. The timeline there is similar in content but presented visually:
Open AI assessed the operation as having no impact or breakout, but rightfully says that it is concerning due to its targeting of a private company.
So, some points here.
Spamouflage is alive and well, and the MPS is presumably hiring local firms to help their efforts.
China is fusing cyber attribution and influence operations in new ways, and they won’t stop targeting US companies, not just “legitimate” IO targets.
They still can’t get their act together regarding mass scale operations, and also still are weirdly focusing on basic development for social listening. I imagine they have issues with data and access to platforms.
AI is a scary industry for China, and they know that they can’t gain supremacy against frontier labs, and it’s best for them to thus limit America’s AI development as much as possible. Surprised we didn’t get amplification of AOC or Bernie Sanders’ insane moratorium bill.
Open AI has an interest in promoting this type of stuff to protect their own interests, but they are also right to expose this and lobby for protection or action.
That’s it for now, thanks for reading.