OpenAI says it has banned ChatGPT accounts linked to an influence operation presented publicly through the International Burke Institute, a self-styled Israel-based think tank.
The company announced the action on October 8 and rated the operation 5/6 on its own impact scale. OpenAI called it “the most far-reaching influence operation it has disrupted,” according to NBC News.
What OpenAI reported
OpenAI said 34 of 36 sampled articles published on the institute's website were copied from elsewhere. Some of those pieces were attributed to authors who did not write them.
The organisation presented itself as an “expert communi...” in the available reporting. The supplied material identifies the operation as Russian and describes the institute as its front.
The findings concern sampled articles rather than every article on the website. further details about the banned accounts, the copied publications or the authors whose names were used is not yet available.
Why the finding matters
The reported action links the use of ChatGPT accounts to an online publishing operation that OpenAI assessed as highly consequential on its own scale. The reported copying and disputed attribution formed part of the company's description of the activity.
No further enforcement measures, legal action or response from the International Burke Institute are included in the available material. The supplied reporting also does not state how long the operation ran or how many people it reached.
Conclusion
OpenAI says it disrupted a Russian influence operation associated with the International Burke Institute, banned linked ChatGPT accounts and found widespread copying among the sampled articles.
Frequently Asked Questions
Q. What did OpenAI ban?
OpenAI said it banned ChatGPT accounts tied to the operation.
Q. Which organisation fronted the operation?
The operation was fronted by the International Burke Institute, described as a self-styled Israel-based think tank.
Q. How many articles were sampled?
OpenAI sampled 36 articles from the institute's website.
Q. How many sampled articles did OpenAI say were copied?
OpenAI said 34 of the 36 sampled articles were copied from elsewhere.
Q. What issue did OpenAI identify with some bylines?
OpenAI said some articles were attributed to authors who did not write them.
Q. How did OpenAI rate the operation?
OpenAI rated it 5/6 on its own impact scale.
