OpenAI disrupts Russian ‘Dark Clark’ and Iranian ‘Bogus Bylines’ false-front influence operations; first Category 5 operation it has found
Confirmed
The takeaway
On Oct 8, 2026 OpenAI reported banning two covert influence operations that used ChatGPT to run “false front” entities.
Status
- Claim
Confirmed
- Our reporting
- High confidence
- Importance
- 3 of 5
- Last verified
- 8 October 2026
Your AI and this story
- GPT-6 Astra161 days after its cutoff
- Claude Opus 5.5100 days after its cutoff
- Gemini 3.8 Flash191 days after its cutoff
- Grok 4.7130 days after its cutoff
None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 100 days before it.
Key facts
- Report: ‘Disrupting AI-enabled false front operations’ (OpenAI, Oct 8, 2026; previewed exclusively by NBC News). Both operations used ChatGPT mostly for drafting internal reports, plus content creation; both exaggerated their effectiveness to their bosses
- Russia, ‘Dark Clark’: ChatGPT accounts operated from Russia via VPN, mostly prompting in Russian, targeting Latin America (undermining Ukraine; also Argentina’s Javier Milei, Bolivia and Ecuador’s Daniel Noboa). They ran a ‘Social Research Center’ through a fake persona ‘Mia Clark’ and appear to have co-opted unwitting local staff. Some claimed fakes match open-source attributions to ‘Politology’/‘La Compania’, a reported Wagner successor
- Examples claimed by the operators and partly corroborated: a fake email from Lima’s Regional Directorate of Education (May 2026) telling schools to hold Ukraine events citing Stepan Bandera, then planted stories in Peruvian and Polish media (a Polish MEP commented); a fake email in June getting Ecuadorian schools to pledge allegiance to Noboa and Erik Prince; fake audio of Ukraine’s consul in Ecuador; fake audio of a La Paz water-company worker during Bolivia’s May protests
- Impact: OpenAI rates Dark Clark Category 5 on the 1–6 IO Breakout Scale (‘the first Category 5 operation we’ve disrupted since we began our reporting’), with evidence it led to public comment by politicians in several countries; information shared with authorities
- Iran, ‘Bogus Bylines’: Persian-language operators via VPN used seven journalist bylines to pitch and refine long-form English articles on the US-Iran conflict; almost 100 articles were published or syndicated across about a dozen small and medium outlets (Feb–July 2026); social-media comments got little engagement. Category 4; appeared to be a commercial for-hire actor
- Pattern: of the 30 covert IO OpenAI has exposed since early 2024, those that placed content in real media outlets reached categories 4–5, while social-media-led ones stayed at 1–3
What happened
OpenAI’s threat-intelligence team published its latest disruption report, focused on two operations that hid behind seemingly independent organizations rather than fake social-media accounts. OpenAI compares them to pre-AI operations such as the GRU’s “Alice Donovan” persona and the IRA-linked “PeaceData” site, and says AI mainly made their workflows faster and more fluent.
Why it matters
It is the highest-impact AI-assisted influence operation OpenAI has publicly attributed, and it shows that the effective use of AI in propaganda so far is in planting content through real institutions (schools, outlets), not in mass posting.
Note: the operators’ own claims of impact are self-reported; OpenAI says it corroborated only some of them with open sources.
Sources
2 sources from 2 sites. Numbers match the chips in the text.
2 sources: 1 primary, 1 press
Primary
- OpenAI: Disrupting AI-enabled ‘false front’ operationsopenai.com, official
Press
Changes
- Filed (OpenAI report read via reader proxy, NBC News)