The New York Times on Wednesday defended an image it published of a massive crowd in Enghelab Square in Tehran, Iran, on Monday after conservative accounts accused the paper of using AI to manipulate it.
An X post from the “Empirical Research and Forecasting Institute” claimed the photo, which showed a crowd celebrating the announcement of Mojtaba Khamenei as Iran’s new supreme leader, showed signs of “digital manipulation.” The characteristics included “a uniform noise texture across the crowd, anomalous clean values around the fountain and suspiciously consistent flag density suggesting copy-paste duplication.”
The claims of AI manipulation were amplified by other conservative accounts, some of which also claimed the photo was from 2020. Posts from Eyal Yakoby, Mark Levin and Sana Ebrahimi claimed the paper’s decision to publish it amounted to a propaganda effort on behalf of the Iranian regime.
But Times spokesperson Nicole Taylor pushed back in a statement to Yakoby, Levin, Ebrahimi and the Institute, saying the photo taken by Arash Khamooshi was “a genuine image” taken on March 9.
“The analysis shared by the Empirical Research and Forecasting Institute is fundamentally flawed: it is dishonestly based on a re-posted version of the originally published image, which misrepresents standard image compression behavior,” Taylor wrote. “The New York Times does not use A.I. to generate or manipulate images to represent real facts; we rely on the work of human beings to bear witness, gather the facts and help readers better understand the world.”
The Times’ principles on generative AI state that the paper must “tell readers how our work was created” if it uses the technology, and the story featuring the image does not include any addendum. The company‘s use of AI has become a flashpoint in its union negotiations over guardrails for its potential future use.
