Identified: AI-Driven 'Iranian Propaganda Campaign' Utilizing ChatGPT Platform
ChatGPT under the Spotlight: Iranian Influence Operation Uncovered
On a Friday shocker, OpenAI admitted to nabbing an influential operation utilizing ChatGPT, dubbed Storm-1679. This cyber group, as reported by OpenAI, initiated a campaign to shape public perception around Vice President Kamala Harris and former President Donald Trump, as well as several other global issues.
Storm-1679, aside from targeting prospective 2024 U.S. presidential candidates, also focused on heated discussions like Israel's Gaza invasion and Olympic participation, U.S.-based Latinx rights, Venezuelan politics, Scottish independence from the U.K, and more.
A majority of the generated content by Storm-1679 went mostly unnoticed, with limited engagement from the general public. Nevertheless, OpenAI took the matter seriously and delved deeper, finding more than a dozen accounts masquerading as both conservatives and progressives. These accounts exploited hashtags like "#DumpTrump" and "#DumpKamala" to further their agenda. Not only on Twitter, but at least one Instagram account was involved in spreading AI-generated content, according to OpenAI.
This is not the first time OpenAI reported state-affiliated threat actors leveraging their tools. However, it's the first time they've revealed a specific election interference campaign involving ChatGPT. Upon discovering the deceit, OpenAI promptly barred a network of accounts responsible for the operation, and shared threat intelligence with a myriad of government, campaign, and industry stakeholders.
Concurrently, a report from Microsoft on August 6 painted a similar picture of Storm-2035 – an Iranian network that fabricated news outlets and posted content centering around election, LGBTQIA+ rights, and Israel's Gaza invasion.
Foreign interference in U.S. elections has become a recurring theme. Previous incidents include Iran-linked phishing attacks and Russia's infamous Guccifer 2.0 hack, which swiped thousands of Democratic National Committee emails and documents in 2016.
Responding to these undercurrents, tech giants have introduced measures such as meme fact checks, political ad bans, war rooms, and collaborations with rivals and law enforcement agencies. Despite these efforts, intricate influence operations like Storm-1679 persist, underscoring the ongoing battle between AI-driven disinformation and the digital watchdogs guarding the net.
Remarkable Insights
- IRANIAN CYBER OPERATIONS: Iran's history includes various cyber operations, including utilization of large language models (LLMs) for phishing attacks and malware evasion. While there is no specific evidence to confirm the involvement of ChatGPT, Iran's cyber tactics are worth noting.
- POTENTIAL TARGETED TOPICS: If an operation like Storm-1679 were linked to Iran, topics such as disinformation, phishing, social media manipulation, and policy dividing subjects could be targeted.
- RESPONSE BY OPENAI AND OTHER STAKEHOLDERS: If a threat is detected, OpenAI might improve monitoring tools, enforce security updates, and collaborate with cybersecurity agencies and platforms to eliminate AI-generated disinformation.
- The future of tech in combating disinformation may be tested as Iran's history of cyber operations, including the use of large language models in phishing attacks, raises concerns.
- As seen with Storm-1679, potential targeted topics in future Iranian cyber operations might include disinformation, phishing, social media manipulation, and issues that polarize society such as the Israel-Gaza conflict.
- In response to such threats, OpenAI, along with other stakeholders, may strengthen their monitoring tools, enforce security updates, and collaborate with cybersecurity agencies and platforms to eliminate AI-generated disinformation.
- The ongoing battle between AI-driven disinformation and digital watchdogs like OpenAI is highlighted by intricate influence operations such as Storm-1679, emphasizing the need for continually evolving strategies in tech against disinformation campaigns.
