Artificial intelligence company OpenAI announced on August 16 that it had disrupted an alleged “covert Iranian influence operation” using its popular ChatGPT software to generate content designed to influence public opinion ahead of the U.S. presidential election.

The company said it had “blocked accounts linked to an Iranian influence operation that uses ChatGPT to generate content on various topics, including the US presidential election campaign.” Although the organization claims the attacks are of Iranian origin, it has not provided any further information to prove the source of the sites, which are not online until May 2024.

However, OpenAI stated that they have seen “no evidence that this content has reached a significant audience.” In other words, the X accounts and websites appear to have few views and are immediately recognizable as AI-authored content.

According to OpenAI, the operation, dubbed “Storm-2035,” used ChatGPT primarily for two purposes: “Creating long-form articles and shorter social media commentary.” The content was then distributed across various social media accounts and websites, including Teorator.com.

A second website mentioned by OpenAI as a potential Iranian influence site is called “savannahtime.com” and was created in July 2024. It targets the liberal end of the US political spectrum with articles through August 18. However, this site appears to have taken its images from other websites, including the Boston Globe. Bne IntelliNews learned.

“We identified a dozen accounts on X and one on Instagram that were involved in this operation. Some of the X accounts posed as progressive, others as conservative,” OpenAI wrote on its blog.

Bne IntelliNews could not verify the reference to the tweets about X, with internal investigations showing that no accounts with the redacted comments reported by OpenAI are currently active.

OpenAI said the impact was limited, although the role of the website and accounts may have been overstated, and IP addresses were not disclosed to ensure the attack originated in Iran.

“The majority of social media posts we identified received few or no likes, shares or comments,” the company said.

OpenAI reiterated its commitment to preventing such abuses, stating: “OpenAI is committed to preventing abuse and improving transparency around AI-generated content. This includes our work to detect and disrupt covert influence operations (IOs) that seek to manipulate public opinion or influence policy outcomes while concealing the true identity or intentions of the actors behind them.”

Tehran’s tech sector is unimpressed

OpenAI’s announcement that Iran was behind political influence campaigns has sparked panic in the Iranian tech community, which already faces hurdles such as a lack of good internet and financial connections to the outside world.

In this latest company announcement, OpenAi did not disclose whether it had shared its user data with Microsoft to determine the origin of the accounts.

A developer in Tehran said upon hearing the news: bne IntelliNews, “This does not bode well. The US is constantly making excuses to disconnect our developers from the World Wide Web,” he added. “My project is mainly based on OpenAi.”

Several tech websites in Iran have taken note of the news with headlines such as “OpenAI blocks Iranian users.” However, sites such as ITAnalyze.ir pointed out that there was no proof that the users were Iranian.

The website wrote on August 17: “These anti-Iranian claims come at a time when Iran’s permanent representative to the United Nations said in an email that the Iranian government had neither the desire nor the motivation to interfere in the U.S. presidential election.”

Iranians form one of the world’s largest groups of software developers and are behind several major technological innovations in Silicon Valley and elsewhere, including helping to found Google and eBay.

Nevertheless, before leaving Iran to seek their fortune, many often struggle with the poor internet connection in the country and live permanently in virtual private networks (VPNs) to conduct business and access the World Wide Web, as many popular websites, including Facebook, Instagram, TikTok and Telegram, are currently blocked by the state in the country.

Previous announcements from Microsoft

Technology giant Microsoft raised the alarm over Iran’s increasing attempts to influence the upcoming U.S. presidential election, according to a report published on August 8.

In the first advisory of its kind, the Microsoft Threat Analysis Centre (MTAC) has observed an increase in activity by groups linked to the Iranian government whose goal is to influence American voters and collect information about political campaigns.

“In recent weeks, groups affiliated with the Iranian government have increased two types of activities,” said Clint Watts, general manager of MTAC.

“First, they laid the groundwork for influence campaigns on current election issues and began to activate those campaigns, apparently in an effort to stir up controversy or influence voters – especially in swing states.”

“Sefid Flood, an Iran-linked influence actor, began preparing influence operations for the 2024 U.S. elections after the Iranian New Year in late March. Sefid Flood specializes in posing as social and political activist groups to a targeted audience in order to foment chaos,” the report said, adding that Iran is undermining trust in the authorities and sowing doubts about the integrity of the elections.