As the Israeli war on Gaza continues, Israeli propaganda campaigns on social media have intensified to influence public opinion and disseminate supportive information for Israeli propaganda. On May 29, Meta, the owner of Facebook and Instagram, removed a network of hundreds of fake accounts linked to an Israeli company. The company, called STOIC and based in Tel Aviv, established a network of artificial intelligence-operated accounts to promote Israeli propaganda and spread misleading claims, especially among the Arab audience.
Simultaneously, following Meta's announcement, OpenAI, the owner of the artificial intelligence software ChatGPT, announced on Thursday, May 30, that it also banned a network of accounts belonging to the same Israeli company. The Israeli network used artificial intelligence models to create articles and comments supportive of Israel across multiple platforms, especially Facebook and Instagram. The company stated that it disabled the activity originating from the network as part of an operation dubbed "Zero Zeno," and some of the fake accounts were created to appear as if they were managed by Jewish students and African-American citizens.
What Is STOIC?
According to its website, STOIC was founded in 2017 and specializes in innovative campaign management, business and data analysis, and providing assistance to clients in making data-driven strategic decisions.
However, STOIC is implicated in the Israeli war on Gaza by disseminating content against Palestinians and other countries such as Qatar. OpenAI stated that the network began its operations in May, issuing automatically generated comments focusing on India and criticizing the ruling "Bharatiya Janata Party" in the Republic of India. It added that it banned a group of accounts belonging to the network operated from Israel, which were used to create and edit content to positively influence the Israeli narrative. The campaign targeted audiences in Canada, the United States, and Israel with content in both English and Hebrew. Additionally, the campaign spread false claims against pro-Palestinian protesters at American universities across various social media platforms.
META, on the other hand, stated that the network, based in Israel, posted comments generated by artificial intelligence praising the Israeli army on the pages of media institutions and public figures. The company said it removed many of these networks before they could build a large audience.
How Did STOIC Promote Pro-Israel Content?
Based on investigations conducted by OpenAI, the company identified a set of trends in how STOIC clandestinely used artificial intelligence models like ChatGPT. These were heavily utilized to generate content, enabling the Israeli company to produce texts (and sometimes images) in larger quantities with fewer linguistic errors than human operators alone could achieve. However, these operations did not exclusively rely on artificial intelligence; the company employed mixed methods. AI-generated content was just one type among several others, including manually written texts or internet-sourced meme images, to make the content appear more human-like. This complexity makes it difficult to detect this type of misleading content.
Israeli Bots on Social Media Platforms
Artificially intelligent bots, known for their automated tasks online, have evolved to engage in content creation and user interaction. These bots possess various capabilities, from streamlining routine tasks to swiftly interacting with users and efficiently managing marketing campaigns. However, their misuse presents dangers such as the dissemination of misinformation, manipulation of opinions, and breaches of privacy. Social media platforms report an ongoing battle against the proliferation of AI-generated fake content, with AI-driven bots being a prominent concern.
In a report from March of 2023, hundreds of social media accounts, actually operated by AI bots, were identified. These bots were observed promoting "Israeli interests" to Western audiences, including American lawmakers, while also spreading unverified promotional content. This included misleading accusations against the United Nations Relief and Works Agency for Palestine Refugees (UNRWA).
Conversely, Lebanese researchers Ralph Baydoun and Michel Semaan have been monitoring "Israeli bots" since October 7, 2023. These bots have been observed disseminating pro-Israeli posts on social media. According to Baydoun and Semaan, almost every tweet from these bots is met with attacks from multiple accounts, all following very similar patterns and appearing almost human-like. However, they are simply bots, not actual humans.
How Do Bot Accounts Operate?
According to Baydoun, the main goal of pro-Israel bots, which have been primarily discovered, is to spread doubt and confusion about the pro-Palestinian narrative rather than gaining trust among social media users. Bot armies, ranging from thousands to millions, are utilized in extensive misinformation campaigns aimed at influencing public opinion. As bots evolve, distinguishing between bot-generated and human-generated content becomes increasingly challenging.
According to an investigation conducted by the Influeanswers Foundation, social media bots have become more sophisticated thanks to artificial intelligence models. These models enable anyone to create simple guidelines for responding to comments, making highly deployable bots relatively easy to spread.
An Increasing Surge of Fake Bots on Social Media Platforms
According to a study by the American cybersecurity company Imperva, malicious bot activity has reached its peak, constituting 34 percent of internet traffic, while the remaining 15 percent is attributed to benign bots.
Step 1: Identifying the Target
Highly advanced bots target valuable users, such as trusted accounts or those with widespread reach, using specific keywords or hashtags. For example, pro-Israel narrative bots monitor accounts using tags against the Israeli narrative, such as #Gaza #Ceasefire.
Step 2: Creating Targeted Responses
Responses to targeted posts are generated by inputting their content into a large language model like Chat GPT.
Step 3: Responding to Posts
Bots can generate responses within seconds, with slight delays added to make them appear more human-like. If the user responds, the bots continue to interact indefinitely.
Several bots can operate from different accounts simultaneously, as evidenced by a recent surge of bots promoting the Israeli narrative, monitored by Misbar. The video shows robotic accounts regularly posting supportive comments for Israel on Instagram.
Artificially intelligent bots can be identified by several distinct signs, according to the InflueAnswers Foundation:
- Profile Picture: The profile picture may lack human features and appear AI-generated or stolen from the internet.
- Name: The name may contain suspicious numbers or follow an irregular format of capital and lowercase letters.
- Personal Information: The bio usually contains minimal personal information.
- Creation Date: The creation date is recent, or the account began its activity recently, which can be observed in accounts that became active simultaneously with the war on Gaza.
- Account Following: Bots often follow each other to increase the number of followers.
Posts: Posts may be entirely random to appear more human-like, such as posting a video clip followed by a completely contradictory post. However, these indicators should be used cautiously, as some genuine accounts may exhibit some of these behaviors.
Other characteristics that can be used to identify bots on social media include:
- Activity: Bots are known for their rapid responses to posts, often within 10 minutes or less.
- Frequency: Due to their lack of need for rest, bots may post on numerous topics, frequently throughout the day.
- Language: Bots may use strangely formal language or exhibit unusual sentence structures.
- Targets: Bots often target specific accounts, such as verified accounts or those with a large number of followers, focusing on significant and controversial topics, such as the misleading campaign against UNRWA.
Since the onset of the Israeli war on the Gaza Strip in October 2023, Israel has employed various human and non-human tactics to propagate its narrative and justify its violent war. The investigation has observed many of these methods, including the enlistment of social media influencers to promote pro-Israel propaganda, the dissemination of AI-generated misleading images, the use of a network of fake accounts, and the deployment of AI-driven bot accounts.
Read More
What Is Behind "All Eyes on Rafah" Slogan That Went Viral on Social Media
The Impact of Instagram’s Political Content Limitation on Public Opinion Trends
