AI-generated disinformation targets COP30 climate summit
AI-generated disinformation about the COP30 climate summit in Brazil circulated widely on social media, including deepfake videos and false documents, according to a report by the Coalition Against Climate Disinformation. The reports found a 267 percent surge in climate-related disinformation from July to September and identified AI-generated content that TikTok did not remove despite being flagged.
What happened
AI-generated disinformation related to COP30, the UN climate summit hosted in Belem, Brazil, circulated widely on social media ahead of the event. The Coalition Against Climate Disinformation and Observatory for Information Integrity documented a 267 percent surge in COP-related disinformation, with more than 14,000 examples recorded from July to September.
One widely-shared video claimed to show massive flooding in Belem but was entirely AI-generated: the reporter, people, flood, and city depicted do not exist. Other videos falsely implied Belem was unfit to host the conference, including footage filmed in Tbilisi, Georgia, and recycled footage from two years prior.
A document claimed to have been written by Grok 3 AI wrongfully dismissed the credibility of the Intergovernmental Panel on Climate Change's climate models.
Reported impact
The disinformation, supercharged by AI and circulating on social media, helped sustain an undercurrent of hostility towards science despite increasing public support for climate action policies. The Coalition Against Climate Disinformation stated that Big Carbon's spending and Big Tech's algorithms are preventing public engagement, exposing people instead to repeated falsehoods.
Index analysis
The video showing the purported flooding did not disclose its use of AI and was flagged to TikTok by Observatory for Information Integrity researchers, but the platform did not remove it. The use of AI-generated synthetic media—including deepfake video—to create climate disinformation reflects a wider trend of AI-tainted climate content throughout 2025.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.