OpenAI flags Iran, Russia AI influence campaigns as high-severity threats
OpenAI rates two foreign influence operations using AI journalists and think tanks as high-impact, marking a first for the company’s reporting standards.
Key points
- OpenAI identified influence campaigns linked to Iran and Russia.
- The campaigns utilized AI-generated journalists and think tanks to sway Western media.
- Both operations received severity ratings of 4 and 5 out of 6, the highest reported by OpenAI.
OpenAI has publicly identified and rated two influence campaigns linked to Iran and Russia as high-severity threats, marking the first time the company has reported what it considers high-impact influence operations. According to CyberScoop, the artificial intelligence developer revealed that these state-linked actors utilized AI-generated journalists and think tanks to attempt to influence Western media outlets.
The campaigns were assigned severity ratings of 4 and 5 out of 6, respectively. These ratings represent the highest level of impact OpenAI has reported to date for influence campaigns. The disclosure highlights a shift in how the company evaluates and communicates the risk posed by coordinated inauthentic behavior leveraging generative AI technologies. By assigning specific numerical severity scores, OpenAI is providing a clearer metric for the potential damage caused by these digital influence efforts.
Context and Background
The identification of these campaigns underscores the growing intersection of state-sponsored disinformation and generative AI tools. According to CyberScoop, the specific tactics involved the creation of AI journalists and the simulation of think tanks. These entities were used as fronts to push narratives designed to shape public opinion and media coverage in Western countries.
This reporting is significant because it establishes a baseline for OpenAI’s threat assessment framework regarding influence operations. Previously, the company had not reported campaigns with this level of severity. The use of AI to mimic credible journalistic sources and academic or policy institutions (think tanks) adds a layer of sophistication to traditional disinformation tactics, making detection and attribution more complex for media organizations and security professionals.
Who Is Affected
The primary targets of these influence campaigns are Western media organizations and the broader public consuming news in those regions. By infiltrating the information ecosystem with AI-generated content that mimics professional journalism and policy analysis, the campaigns aim to distort public discourse. Media professionals, editors, and fact-checkers are directly affected, as they must now contend with highly realistic synthetic content designed to bypass traditional verification processes.
Security and technology professionals monitoring threat intelligence feeds are also affected, as this disclosure provides new indicators of compromise and tactical patterns associated with Iranian and Russian state actors. The high severity ratings signal that these campaigns are not minor nuisance activities but are considered significant threats to information integrity by the platform hosting the underlying AI models.
What should you do?
No specific patch or technical fix has been confirmed for these influence campaigns, as they represent coordinated social engineering and disinformation efforts rather than a single software vulnerability. Security professionals and media organizations should remain vigilant for content that appears to originate from unfamiliar journalists or newly formed think tanks, particularly when the narratives align with known geopolitical interests of Iran or Russia.
Organizations should review their content verification protocols to ensure they can identify synthetic media and AI-generated text. Monitoring threat intelligence reports from platforms like OpenAI can provide early warnings of emerging influence tactics. Until further guidance is issued, the best defense is heightened scrutiny of sources and cross-referencing claims with established, reputable news outlets.
Background: Generative AI
Generative AI creates content from prompts. Agentic AI performs actions using tools. The former outputs text or code; the latter changes system states. Understanding this distinction is critical for security architecture, as agentic systems introduce execution risks that pure generation models do not possess.
Read the full guide: Generative AI vs Agentic AI: Architectural Differences and Security Risks
What to do and how to stay safe: OpenAI
- Urgency and secrecy are the primary psychological triggers used to bypass rational decision-making processes.
- Technical indicators like spoofed addresses are often invisible without specific tooling, so behavioral context matters more.
- Verification must occur through a pre-established out-of-band channel, not the one currently in use.
Trust but verify is the only reliable defense against social engineering. Always confirm urgent requests through a separate, pre-established communication channel.
Step-by-step guide: Social Engineering Warning Signs and How to Respond
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
Which countries were linked to the high-severity AI influence campaigns?
OpenAI linked the influence campaigns to Iran and Russia, according to CyberScoop.
How severe did OpenAI rate these influence campaigns?
The campaigns were rated 4 and 5 out of 6 for severity, which OpenAI considers high-impact.
What tactics were used in these influence campaigns?
The campaigns used AI-generated journalists and think tanks to attempt to influence Western media.
