Seated at a desk, the girl smiles, then looks down and begins to write in a notebook. It’s unmistakably Anne Frank. But wait – there is only one known film of Anne and her sister Margot. And this isn’t it.
The new 23-second clip of perhaps the world’s best-known victim of the Holocaust was created using artificial intelligence, part of an educational campaign by the World Jewish Congress.
“It is so incredible that they can animate these photos,” writes the daughter of a Holocaust survivor under the post, which since its release in June 2025 has garnered nearly 800 comments, 16,000 likes, hearts and hugs and 2,000 shares.
“For decades and decades, we’ve seen her photos. Say what you will about AI, it stirs me to see her in ‘real life,’” wrote another commenter, prompting a response: “The same for me, brings her more vividly alive and poignant.”
“Disgusting AI,” retorted another user. “Anne Frank deserves better.”
Stories designed to farm engagement, denounced as a purveyor of “AI slop”
The comments illustrate a divide on the use of AI for Holocaust remembrance and education. Some employ these tools respectfully and responsibly, creating new means to share the stories of victims or of the rapidly diminishing number of living witnesses. At the same time, an ever-growing supply of fake Holocaust posts are designed to generate clicks for their mostly anonymous creators – and run the risk of undercutting trust in Holocaust content at a time when denial is surging.
That risk has some concerned that even well-intentioned Holocaust AI efforts, such as the WJC’s, could end up furthering mistrust in Holocaust documentation and should be avoided.
Now, educational and memorial establishments increasingly are asking: Does AI help preserve memory, or does it contribute to Holocaust distortion and denial? Does it mobilize the public to learn from the past, or contribute to a world where people can’t know what is true anymore? Does it humanize dry facts and figures, or does it erase something that at the heart of what makes us all human?
The answers to all are “yes,” said Yfat Barak-Cheney, executive director of the WJC’s Institute of Technology and Human Rights, who last month held an invitation-only “fireside chat” with Meta’s public policy director, Ben Good, discussing some of those very questions. According to WJC, Good assured the audience gathered at New York’s Yeshiva University that the parent company of Facebook and Instagram is committed to fighting antisemitism on its platform, including amid the rise of AI-generated content.
While AI opens many doors, it also “is a convenient tool for Holocaust distortion,” Barak-Cheney told JTA before the event. She acknowledged that “there’s a huge ethical question here” on how far to go with AI but added that there is no possibility of avoiding the increasingly ubiquitous tool – which she said has enormous potential for the field of Holocaust remembrance.
“If we’re not going to use any of these tools, we’re going to remain way behind,” she said.
The boundaries were tested recently in the creation of a project called “Tell Me, Inge…” by partners Meta, StoryFile, the Claims Conference and UNESCO.
In the project, German survivor Inge Auerbacher answers questions for a hologram interview that will outlive her. Drawing from Auerbacher’s oral history, the project uses AI to find the most appropriate answers for a viewer’s questions. Viewers can interact with a virtual image of Auerbacher using a headset on a desktop or mobile device.
While the answers are generated in part by AI, the creative team decided to stick with the languages Auerbacher knows – German and English – rather than use the tool to “make her seem like she is speaking in any other language,” Barak-Cheney said. Otherwise, “down the line someone could call the whole interview fake. And then we’ve caused damage, instead of assisting with preserving her testimony.”
Robert Williams, CEO and Finci-Viterbi Chair of the USC Shoah Foundation, said “you have to be extraordinarily careful” when using AI for memory projects like these.
“You want to look at the source in whole, in order to try to understand that these are not just moments in time [and] that there’s a context in which these things appear,” Williams said. The hologram interview, he said, “is one way that you can use AI responsibly to get people to engage with the testimony of Holocaust survivors in a way different from just plunking somebody down in front of a computer and saying, ‘Watch.’”
Yet the foundation responsible for stewarding the legacy of Anne Frank cautions that AI images can lead to distrust of the historical record.
“This ‘footage’ of Anne Frank, and anybody else, is going to stay in the ’net,” said Yves Kugelmann, a board member of the Switzerland-based Anne Frank Fonds and the co-producer of a 2021 film based on Frank’s diary using conventional animation. “And in 10 years, you won’t know what is true, what is not true, what footage is correct or which one is an invention.”
Indeed, the growth of AI has led to a mushrooming of outright fake Holocaust posts on social media: completely invented biographies, or real names paired with AI-generated photos. A user who looks at one post will often soon see more, thanks to the algorithms that pick up on their interests – potentially flooding their feeds with Holocaust “slop,” the term for low-quality online content.
One such post that went viral depicts “Hershel Rubin,” who supposedly died in Treblinka. He “liked feeding the family goat,” says the text beneath an angelic face. No victim of that name exists in the Yad Vashem Shoah Names database.
The image was posted on a Facebook page by an entity called “Timeless Tales,” with no other identifying information. Some of the comments on the page, which posts a diverse array of heart-rending stories designed to farm engagement, denounced it as a purveyor of “AI slop.”
Another Facebook page includes an idealized image of a little girl identified as “Lída Kohnová” who purportedly died in Auschwitz. “Who could have imagined that this little girl, racing the wind with flushed cheeks and wild dreams, would one day vanish into the darkest chapter of the 20th century?” reads one of the posts.
Again, no such person exists in the Yad Vashem database.
The image itself includes various hallmarks of AI-generated photos: unnatural or inconsistent lighting, anachronistic details and suspiciously smooth focus. It is posted on a page, “History Pictures,” that has no other identifying information and likewise posts a steady stream of vignettes about the past, accompanied by a mixture of real, AI-enhanced and completely fabricated photographs.
Last year, the Auschwitz Memorial Museum – after hearing from numerous followers about such posts – published a warning about AI Holocaust posts.
“The use of artificial intelligence to generate fictional images of Auschwitz victims … is not a tribute,” it said. “It is a profound act of disrespect to the memory of those who suffered and were murdered in Auschwitz. It undermines the integrity of historical truth.”
Memorial spokesperson Pawel Sawicki asked Meta, the parent company of Facebook, to intervene. Today, some of the offending posts are gone. But new ones have taken their place: Sawicki recently posted a sample of fake images of the death camp’s main gate to show how social media platforms “are contributing to the spread of historical distortion.”
This modern-day Hydra – the more you cut, the more heads it grows – has troubled educators and remembrance institutions.
This past January, Germany’s government and dozens of Holocaust memorial institutions in the country published an open letter calling on social media platforms to do more to fight the spread of “AI slop.”
Truth does not matter to these slop mills, or “content farms,” the letter lamented. Only emotions do.
For example, “AI-generated images are circulating that show an alleged reunion between prisoners and liberators, or fictional scenes of crying children behind barbed wire.” The content is “made up of fragments of historical facts and emotionalized fiction.”
The letter came the same month that educators and eyewitnesses gathered in Berlin at a conference titled “Digital Horizons – The Transformation of the Culture of Remembrance,” hosted by the Central Council of Jews in Germany. Participants weighed the pros and cons of, for example, holograms and virtual reality headsets in Holocaust education.
Some platforms are taking steps to make sure users can access accurate, human-generated information about the Holocaust. In February, Meta launched a feature on Instagram and Threads – already in place on Facebook and TikTok – that steers users searching for Holocaust-related terms to the aboutholocaust.org platform, developed by the WJC in partnership with UNESCO.
Other platforms have managed the threat in other ways. “YouTube in particular had and still has a pretty robust system for addressing problematic content in ways that push it down in the algorithm,” said Williams of the Shoah Foundation. “Meta seems to have a fairly robust internal mechanism.”
But such safeguards are not universal, he warned. “I know that some are more willing than others, and we’re trying to figure out a way to be a resource for them all.”
With every new technology comes risks and challenges. Despite the drawbacks, AI can be a force for good, says Cory Weiss, WJC’s executive director of communications strategy. When working with colleagues on the Anne Frank animated clip and accompanying text, released on her birthday last June 12, he knew that to make the images speak or to change their context “would be crossing a line.”
“For us, it was about getting social media users to pause, in an era when people are very quick to scroll through,” he said. The animated clip was a “one-off shot,” using images that are in the public domain and deliberately not putting words into Anne Frank’s mouth.
He’s encouraged that social media giants are trying to keep an eye out for “false narratives.” Sure, someone unscrupulous could animate an image of Anne Frank, saying, “I was never a real person and everything you’ve been told is a lie,” said Weiss. “But I think most people are smart enough to know that that would be a fake.”
About AI, he said, “Our goal is to try to use it responsibly when we do use it and not go over the top in using it.”
The WJC is, apparently, on the cutting edge of a trend. This week, a new traveling exhibition is opening in Germany with an internet component that includes AI animations of pre-war Jews.
“Living Traces – Discovering Jewish Life” presents portraits of 12 individuals from Fulda, a city in the state of Hesse. The brief animations are based on historical photos.
“The respectful and careful use of AI helped create something very special,” said Anja Listmann, Fulda’s commissioner for Jewish life, who came up with the idea. “Old photos were brought to life with modern technology so visitors can feel a real connection.”
Recently, Weiss – who also manages the WJC’s #WeRemember campaign for International Holocaust Remembrance Day – himself was touched by an artistic Holocaust-related AI animation.
The Instagram post shows ghostlike figures arising from an actual Budapest memorial, “Shoes on the Danube Bank.” The memorial’s cast-iron boots and shoes represent the thousands of Jews who were lined up and shot at the river’s edge in late 1944 and early 1945.
“I saw those animating into children quite literally standing on the banks of the Danube, staring at the river,” Weiss said. “It’s evocative.
“I’m not prepared to say whether I would put a full-on kosher stamp on it,” he added. “I need to watch it a bunch of times and ask, ‘How does this feel?’ And I would want to seek some other perspectives, particularly from the Jewish community there. But it got me to stop and say, ‘I’m not sure this is right, but also it definitely gives you a feeling.’”
