Why We Shouldn’t Be Surprised about #AI #Auschwitz, and What We Can Do About It
By Prof. Victoria Grace Richardson-Walden
Last week the BBC reported on spammers sharing ‘AI slop’ images of Auschwitz for profit. In this week’s blog, we respond to their article, reflecting on our years of research into Holocaust memory and AI.
On Friday 29th August, the BBC published an article with the headline: ‘BBC reveals web of spammers profiting from AI Holocaust images.’ Whilst this might seem shocking, it shouldn’t be, and far from ‘spamming’, which can be considered a flippant action, the individuals involved in these profit-making schemes are using social media in conscious and sophisticated ways – as its logics intend it to be used. This should make us stop and ask: are platforms like Facebook good for Holocaust memory?
On one hand, it would be easy (especially in the current political climate) to say, ‘not really’. On the other hand, if Holocaust museums and educational organisations want their authority to remain recognised widely, they need to be not only be present on these sites but enhance their visibility. As our previous research has argued (Walden 2021), professional Holocaust organisations need to adopt the strategies of apparent ‘bad actors’ if they want their messages to circulate as much as alternative ones.
Let us take a closer look at some of the criticisms levelled in the article.
Social Media as Some Kind of Strange Emotional Game
The BBC article quotes Paweł Sawicki (of the Auschwitz Memorial Museum) as referring to these so-called ‘AI spammers’ as turning the atrocity of the Holocaust ‘into some kind of strange emotional game’. Yet, his words here describe the fundamental logics of these platforms to a tee – whilst the Holocaust may not be a game, the phrase ‘some kind of strange emotional game’ is a fitting description of social media platforms.
‘Affective economies’ is a term popularised by Cultural Theorist Sara Ahmed (2004), referring to the fact that emotions are not simply feelings felt within our individual bodies, by circulating affects that create boundaries and align individuals with others through intense senses of shared feelings (not just beliefs). Since, her seminal work, many scholars have gone on to highlight that ‘affective economies’, and not just ‘attention economies’ as is more commonly discussed, underpin the logics of social media platforms, including the data-mining involved in sentiment analysis and predictive analytics, which are fundamental to the algorithmic arrangement of content (Andrejevic 2011).
These affects are always deeply tied to ideology, as in Ahmed’s work, and intense feelings can led to a belief that what is seen should be accepted as ‘common sense’ and thus not questioned (Pedwell 2024) such as happens with conspiracy theories, including Holocaust denial and distortion (Grandinetti 2022).
The corporations behind these platforms (such as Meta) and those whose income relies on the monetisation of content make their profit by playing a game of affect. That is, the platforms rely on content that encourages emotional responses from the users who view it, which compel them in turn to share, comment, and contribute. Their responses then trigger more responses, thus the affect compounds and spreads virally. The commodification of our emotions is the name of ‘the strange kind of’ social media game and the circulation of provocative content is the core game mechanic.
Social media are a complex web of users, platforms and algorithms, each of which is complicit in this game – but certain individuals working for the platforms are the only ones who have full access to the rule book. This means of course that Holocaust museums who use social media platforms are also complicit in these affective economies, as are researchers who uncritically make use of tools like so-called ‘sentiment analysis’ to assess how users engage with content. Yet, the conundrum remains – to be visible to publics we assume we need to be on social media.
Holocaust Education As A Victim Of Its Own Success
This poses a specific challenge for Holocaust memory, for amidst all the debates about whether Holocaust education has failed or not, what it has achieved is a wide recognition at least that the Holocaust is (a) an important historical event, (b) an emotional one, and (c) increasingly, and concerningly a debatable one (and it is this last point that fuels the ‘failure’ arguments). This of course makes it ripe for virality online – significant and negative events get reactions (we know this even from the ‘legacy’ media days through the concept of News Values (Galtung and Ruge 1965).
In some ways then it is less the failure of Holocaust education and more its success in drawing attention to this particular historical period which has encouraged those who make a living off monetising social media to create Holocaust-related content. As you’ll see in the BBC article, one of the interviewed ‘spammers’ says that history sells, and the Holocaust is one of a few popular topics that appear when they use Meta’s professional dashboard. This is not the first time that the known impact of ‘the Holocaust’ or ‘Auschwitz’ in web culture has been noted. Tomasz Łysak (2021) identified YouTube vloggers from a variety of backgrounds doing their ‘#Auschwitz’ video to get more followers. Holocaust memory and education then have become victims of their own success.
For the ‘spammers’, it is not the subject of ‘the Holocaust’ that matters, but its ability to pull on heart strings or stir debates. Indeed, our research has demonstrated that posts by Holocaust museums that feature children or stories of national heroes do far better than other content: so professional Holocaust organisations also play the affective game (Walden and Makhortykh 2023).
Gen-AI text-to-image translators not only enable much faster production of such content (and posts which are inaccurate) but have also encouraged the Auschwitz Memorial Museum and its followers to reshare these posts exclaiming their outrage. Indeed, the museum has cultivated, rather uniquely, a ‘virtual memory community’ via its social media channels – with bastons of its memory speaking up for the museum and the victims it represents (Dalziel 2021). This resharing of course perpetuates the popularity of the topic on the Meta dashboard and thus reinforces ‘Holocaust’ as a profitable subject matter.
‘Auschwitz’ in particular has long served as a metonym for the ‘Holocaust’ in public imagination. The Auschwitz Memorial Museum is one of a handful of Holocaust-related institutions that has far more reach than the majority as both a memory institution and in terms of its core subject matter – the historical concentration and death camp Auschwitz (Ben-David, Meyers and Neiger 2024; Walden and Makhortykh 2023; Eva Pfanzelter 2017). This demonstrates that the historical ‘canon’ related to Holocaust knowledge is retained on social media rather than resisted by any possible diversity of content. On one hand, a wider diversity of content would diffuse ‘the Holocaust’ into a myriad of different narrative and encounters with the past, creating a less coherent ‘past’ to confront. On the other hand, it would make this past less popular in the dashboard rankings and therefore less ripe for mass GenAI posts whilst exposing publics to a wider variety of places and stories.
The Challenges of Researching Social Media
Empirical research into whether these posts would have been so widely circulated without the public debate instigated by the Auschwitz Memorial Museum is increasingly difficult without the cooperation of tech corporations like Meta, Bytedance and X. As is studying the effect of algorithmic organisation on their visibility (Metzler and Garcia 2023). Since Trump’s inauguration most social media companies have doubled down on their closedness making it more challenging to perform thorough research. Having once been more open to working with humanitarian and Holocaust-related organisations, their priorities have shifted; API access have been closed or adjusted; and content moderation limited. It is worth noting that previous functionalities in all these respects was already problematic but now it is much worse.
Platformization and the Logics of Social Media
We are facing an age of increasing platformization of Holocaust memory, informed by monetization models which have sparked some critics to refer to it as ‘platform capitalism’ (Nick Srnicek 2016). (For more on platformization and memory, see the ongoing collection in Memory, Mind and Media guest edited by Rik Smit.) This particular form of capitalism relies on affective economies to encourage user interaction, shaped by algorithmic arrangement and extraction of data. Algorithmic arrangement has always been a part of social media logics. However, more recently, some have argued that these platforms are no longer ‘social’ media but more so ‘algorithmic media’, with TikTok leading the way in deprioritising social functions (Liang 2022). Last year, Mark Zuckerberg – founder and CEO of Facebook/Meta, declared AI-Generated posts as ‘the likely next wave of content for platforms like Facebook and Instagram’.
Image from Auschwitz Memorial’s Facebook post. Added under fairuse policy.
The logics of platformization, affective economies, algorithmic arrangement and data extraction as we see them on social media today are amplifications of existing mechanisms which already existed – albeit to a much lesser extent – with legacy media, such as television, advertising and film (take for example film and television industry oligopolies, manipulation techniques in advertising, and audience/market segmentation strategies).
On social media and other digital contexts, these logics work together towards another, that is: personalisation. Our media engagement is becoming increasing personalisation, and this further complicates how easy it is to identify – as researchers – to whom the Auschwitz AI content is visible and what they do with this content in their private communication spaces.
Holocaust Studies’ Latest Authenticity Debate
The BBC article also raises the issue of these AI-generated images as ‘fake’ thus it seems Holocaust memory is facing another ‘crisis of authenticity’. But this is not the first and will probably not be the final media panic of this kind. AI has also long been integral to social media platforms. Similar authenticity debates were raised about the trivialisation offered by NBC’s Holocaust and the tasteless adverts that featured during the commercial break (Wiesel 1978; Shandler 1999), or criticism of the mainstream Hollywood melodrama Schindler’s List not being the long-form documentary Shoah (see Hansen 1996 for a debate of this critique). We have also seen purposeful fakeries, such as Binjamin Wilkomirski’s Fragments (non)testimony (Vivian 2017), as well as an abundance of fictional depictions in other formats, some criticised by the Auschwitz Memorial previously: e.g., the Tattooist of Auschwitz novels, and Hunters TV series.
In emerging research, I have been arguing that we need to recognise that digital media are essentially non-representational and ask what their mathesis logics can do for Holocaust memory in order to fully appreciate the extent to which they may (or may not) contribute productively to this cause. To continually judge them as ‘inauthentic’ is to miss the point of what they are and how they function. However, the struggle remains, that others adopt them to create seemingly ‘representational’ images.
What To Do About This
The question is then how can Holocaust memorial museums continue to be visible authorities over this past in this increasingly digital ecology?
What is their purpose and role? How do they perform this? And importantly, where – at physical sites, across online platforms, using algorithms and AI models or not? Where do they draw ethical lines regarding the extent to which they want to work with commercial, corporate platforms?
It is not enough, as suggested in the BBC article that ‘government and philanthropic investment’ has gone towards ‘awareness campaigns’ – one off, or even repeated campaigns such as #WeRemember are not sufficient (see Richardson-Walden and Marrison 2024). Digital capacities need to be taken seriously within the Holocaust remembrance and education sectors. Governments and philanthropic investment needs to be directed towards more long-term, strategic and sustainable digital expertise within and across Holocaust organisations.
We have written a variety of recommendations on using AI, and social media for Holocaust memory and education, directed at policymakers, academics and Holocaust museum professionals (see a list of recommended reading at the end of this blog).
If Holocaust organisations want to use social media they need to engage with the full spectrum of their logics otherwise why invest time and energy into this medium? That said, this is not a rallying call in defence of platformization. The other option is to redirect efforts currently given to social media engagement elsewhere – where might that elsewhere be that would have significant reach?
The challenge for Holocaust organisations is if they want to adopt the logics of social media platforms this means using the strategies of apparent ‘bad actors’ to their advantage. This would involve thinking and strategically planning social media work as if they are commercial entities and working collectively at a global scale. Some techniques would include:
- Making use of sentiment analysis and detailed, regular assessment of trends
- Experimenting with emotionally-impactful content
- Creating matrix collectives that strategically share, post and guest appear in each other’s content (see Richardson-Walden and Marrison 2024)
- Posting at scale (the ‘spamming’ the BBC nods too)
However, most Holocaust organisations, beyond the few majors, are small outfits often relying on volunteers and interns. As our recommendations and policy guidance have emphasised, to move towards even a level playing field with their competitors (e.g., the for-profit entities the BBC identifies), they need substantial resource to support digital strategy.
This means more than simply funding for one-off or recurring campaigns, but support for a systematic network of professional Holocaust memory social media and AI professionals, trained in both the digital expertise and Holocaust-related pedagogy and ethics needed for the job. This would need to be a network that regular collaborated and connected, working together to track broader activity on social media, responding and reacting to content deemed problematic that goes viral, and being strategic in content produced.
It doesn’t really matter whether fictitious content online is produced by AI models or not – but what they add is the speed at which posts can be produced, thus the likelihood of any one post or indeed a series of posts going viral is higher.
Nevertheless, this particular case – deemed headline worthy in the UK – is illustrative of the urgent need for Governments and funders to invest in digital capacities at both the level of transnational organisations like the International Holocaust Remembrance Alliance and in a joined-up way across global Holocaust museums. And also, there needs to be serious consideration about the need for legislating further against the significance control social media corporations have over knowledge production.
We are currently carrying out a survey of global digital Holocaust memory practice, including questions regarding social media and AI practice by Holocaust museums, memorials, and archives. If your organisations has not completed the survey, please do so via this link.
Read our various recommendations and related research here:
Recommendation and Guidelines Reports
Academic Publications
Imagining Human-AI Memory Symbiosis (2024)
An Entangled Memoryscape: Holocaust Memory on Social Media (2024)
Understanding Holocaust Memory and Education in the Digital Age: Before and After Covid-19 (2021)
Watch Our Recent Roundtable on Social Media and the Future of Holocaust Memory
What is the Future of Holocaust Memory on Social Media? (2025)
For Further Research into Social Media and Holocaust Memory
Check out our Zotero bibliography here
Banner image created by the author in MidJourney using the prompt ‘create an image of Auschwitz that looks AI generated’.