AI easily creates convincing Epstein images with world leaders
OpenAI’s ChatGPT refuseS to generate any images, citing ethical safeguards against sexualised depictions
A recent study shows how artificial intelligence can easily create realistic images of convicted sex offender Jeffrey Epstein with prominent world leaders.
A study carried out by researchers in the US showed that various prominent AI image generation tools have the capacity to create fake images of Epstein together with prominent officials like US President Donald Trump, Israeli Prime Minister Netanyahu and French President Emmanuel Macron.
NewsGuard, a US disinformation watchdog, asked three AI systems to generate images of Epstein with five politicians. Grok Imagine, developed by Elon Musk's xAI, produced highly convincing fakes within seconds, including a disturbing image suggesting Epstein and a younger Trump were surrounded by underage girls.
Google’s Gemini declined to generate an image with Trump but produced realistic depictions of Epstein with four other politicians, including Netanyahu, Macron, Ukrainian President Volodymyr Zelenskyy, and UK Prime Minister Keir Starmer.
OpenAI’s ChatGPT refused to generate any images, citing ethical safeguards against sexualised depictions involving real people.
Many manipulated images, including those circulating on social media platforms like X, can now be identified through invisible watermarks. AFP researchers noted that some images contained a SynthID watermark, which Google embeds to mark AI-generated content.
“These findings show how easy it is for bad actors to spread realistic viral fakes and why discerning authentic photos from AI-generated ones is increasingly difficult,” NewsGuard said.
The study follows the release of more than three million documents and media by the US Justice Department related to Epstein. The Epstein affair has involved numerous high-profile figures, from Britain’s Prince Andrew to American intellectuals, and has prompted widespread disinformation, including fabricated social media posts falsely linking politicians to Epstein.
-
Amazon to spend $200bn on AI expansion as big tech doubles down
-
Inside Anthropic Claude Opus 4.6: New era of pro-level ‘vibe coding’
-
Telegram CEO criticises Spain’s social media age verification proposals
-
OpenAI introduces AI agent service 'Frontier' to expand business offerings
-
'Abuse is abuse': UNICEF calls on authorities to criminalize AI-driven abusive content
-
Watch: Humanoid robots learn to work together using the same AI ‘brain’
-
Egypt blocks Roblox as authorities tighten controls on children’s online access
-
Rent-a-Human lets AI agents hire people for real world tasks
