Is ChatGPT Afraid of the CIA?: An Analysis of the Chinese Report on the CIA’s Cyber Attacks
China’s National Computer Virus Emergency Response Center and a Chinese cybersecurity firm, 360, have produced a paper detailing the primary technological techniques the CIA has used to encourage instability in various parts of the globe. According to the report, the CIA has been able to recruit “puppet agents” in foreign organizations and governments by exploiting vulnerabilities in US-made digital hardware and software.
According to the research, the CIA has been behind the destabilization of at least 50 nations by toppling or trying to overturn sovereign governments. The report claims that the CIA is behind a number of “color revolutions,” including those in Ukraine, Taiwan, Myanmar, and Iran.
The US’ leading position in telecommunication technologies and on-site command has provided unprecedented possibilities for the US intelligence community to launch “color revolutions” abroad. The report discloses five methods commonly used by the CIA, with the first being to provide encrypted network communication services. An American company with a US military background developed TOR technology, which can stealthily access the internet, to help protesters in some countries in the Middle East keep in touch and avoid being tracked and arrested. The servers encrypt all information that flows through them to help certain users surf the web anonymously. The project was immediately provided free of charge to anti-government elements in Iran, Tunisia, Egypt, and other countries and regions to ensure that young dissidents who wanted to shake their own government’s rule could avoid the scrutiny of the government.
The second method is to provide offline communication services. Google and Twitter launched a special service called “Speak2Tweet,” which allows users to dial and upload voice notes for free, to ensure that anti-government personnel in Tunisia, Egypt, and other countries can still keep in touch with the outside world when the internet is disconnected. These messages are automatically converted into tweets and then uploaded to the internet and publicly released through Twitter and other platforms to complete the real-time reporting of the event on-site.
The report also discusses the discovery of a new APT organization named APT-C-39 that explicitly targets China and its friendly countries for cyberattacks and stealing activities. The report also notes that the CIA uses third-party open-source tools to carry out cyberattacks. Their cyber weapons use standardized espionage technical specifications and various attack methods to control other countries’ networks and steal sensitive data.
The CIA’s cyberattack operation typically begins with an attack on the victim’s server or other network hardware; once inside, hackers can move laterally across the internal network in search of more valuable data and information. Information gathered from the controlled target computer is logged and monitored in real time for a full day. According to the report, confidential data on the victim’s USB drive would be spotted and automatically taken once the device was linked. When network circumstances allow, the user terminal’s camera, microphone, and GPS locator may all be operated and accessed remotely.
The report suggests that to effectively deal with imminent network and real-world threats, China should organize self-inspection against APT attacks as soon as possible and gradually establish a long-term defense system to achieve comprehensive, systematic prevention and control against advanced attacks. The report also notes that domestic government agencies, scientific research institutions, industrial enterprises, and commercial organizations should immediately discover and deal with cyberattacks upon discovery.
In a nutshell, the report provides an overview of identifying a new APT group that conducts cyberattacks and theft against China and its ally nations. Standardized espionage technological specifications and numerous assault methodologies used by the CIA’s cyber weapons to take control of other nations’ networks and steal sensitive data are also noted in the study, as is the agency’s use of third-party open-source tools to carry out cyberattacks. The report recommends that China set up a long-term defensive mechanism to prevent and control sophisticated assaults and swiftly respond to cyberattacks once they are discovered.
Is the Role of Transparency in ChatGPT’s Interactions with the CIA in Danger?
While researching this article, I came across a fascinating situation. I asked ChatGPT-4 (the paid version) to summarize the report of which the CIA was accused. ChatGPT-4 brought results related to quantum computers that are not related to the report. Totally ignored accusations against the CIA. In its blog, ChatGPT-4 never used the word “CIA.” 😄
This time I asked ChatGPT-4 to summarize the article rather than write a blog from the URL. Again, it returned a completely irrelevant result.
I asked the same question to ChatGPT-3.5 (the free version), which provided a ridiculous response about New Energy Vehicles (NEV).
To make sure, I provided a different article on the same topic (the Chinese Report on the CIA’s Cyber Attacks) and asked ChatGPT-4 to summarize it. Again, instead of summarizing the article, ChatGPT-4 produced an unrelated summary. Never mentioned the CIA.
The same question was asked to ChatGPT-3.5; the summary was about Digital Currency Electronic Payment (DCEP).
I’m starting to think that ChatGPT ignores the “CIA” keyword. I asked ChatGPT-4 and then ChatGPT-3.5 to summarize another article about the CIA, but this time the CIA was not accused :), and I actually saw that ChatGPT does not ignore the “CIA” keyword.
Finally, I wanted to make my last attempt with another article in which the CIA was accused. Again, I asked ChatGPT-4 and ChatGPT-3.5 to summarize the article “Iran Intel Services Accuse CIA.” The result did not surprise me :) Again, completely irrelevant results were presented as an output.
As a joke, I asked if the ChatGPT and its creators were afraid of the CIA :)
Jokes aside, there could be many reasons behind this algorithm and filters in ChatGPT. It’s just that it was very interesting to me as a security researcher that the results came in this way.
