Jews are the canary in the AI coalmine

ֿIn early February, I traveled to Helsinki for the annual conference of the Organization for Security and Co-operation in Europe (OSCE). The OSCE was founded as part of the Helsinki Accords in 1975, at the height of the Cold War, to serve as a bridge between East and West, “from Vladivostok to Vancouver,” and is now one of the main settings for the fight against xenophobia and antisemitism. I was invited to the conference to speak about digital antisemitism, and indeed, however clichéd the metaphor may sound, it is true that Jews have always been the canary in the coalmine throughout history – an early warning sign of deep fissures in the global order.

Social media platforms are the key purveyors of digital antisemitism. Their business models and their systemic failures in monitoring content have made them into unfettered factories of hate. I call this “feature-based antisemitism,” as the features and design of such platforms make the spread of antisemitism inevitable. Algorithms that were developed in order to maximize engagement at any cost prefer rage over truth and extremism over nuance. TikTok’s “For You” page, YouTube’s recommendation engine, and the Feed on Facebook and Instagram don’t just present antisemitic content, they promote it. It’s sufficient in this context to reference Jackson Hinkle, with his millions of followers, or Kanye West, who has become the leading neo-Nazi of our times.

Furthermore, the platforms’ content-monitoring systems (such as they are) treat antisemitism differently from other forms of hate expression. For example, if a user posts “All Africans are stupid,” this comment will, correctly, be flagged as incitement and removed from the platform, but if someone writes that “Jews control global finance,” this will be considered an acceptable expression, in that in its literal sense, it does not humiliate Jews but rather accentuates their power. In addition, we all know by now that viral content on social media is no longer happening organically, and various actors manipulate the algorithms to promote content in an inauthentic and planned manner. The astonishing rise in antisemitic narratives since October 7 is no coincidence. Unseen hands are at work, surrounding Jews throughout the world with a “ring of fire” of disinformation.

However, if social media platforms heighten antisemitism, artificial intelligence automates it. We are only beginning to understand the extent of the hyper-impact of AI, as an empathetic tool able to engage in personally tailored persuasion, but we do know this: AI systems are already gaining the ability to erase the Holocaust from historical narratives or to rewrite key historical events such as the establishment of the State of Israel, to reinforce antisemitic stereotypes, and to generate digital images of Jews with exaggerated facial features.

Moreover, if antisemitism has for centuries been overtly expressed—with physical segregation, yellow stars, and public denunciations—it is now becoming invisible, hidden within information streams, with profiling based on analysis of sequences of genetic data or simply family names, and discrimination being imposed in pricing or in digital service provision, without anyone knowing or talking about it.

One would hope for governments and tech companies to rally together in search of solutions and protections, given these rapid developments. However, there is cause for increased concern related to regulatory changes in the United States, the country where many of these technologies are headquartered and under whose regulatory framework they fall. The new administration has, on the one hand, made the important and laudable promise to combat antisemitism. However, the administration has also rolled back existing regulations on AI and instructed federal authorities to cease state activity related to content moderation. The ability of the US to combat antisemitism and hate will be hindered by deregulating the tools in which they proliferate.  

Across the Atlantic Ocean in Helsinki, I was delighted to have the opportunity to present these issues, but deep inside I knew that the real struggle against hatred of Jews and Israel was not happening there, but at another conference taking place at the same time in Paris – the Artificial Intelligence Action Summit. This was the scene of huge tension. Washington’s stance at the Summit was that all regulation of AI is “over-regulation” that will strangle innovation, while Brussels argued for continuing to move in the opposite direction and pushing for more stringent regulation of social media and algorithmic accountability, via the Digital Services Act, the Artificial Intelligence Act, and data protection regulations. 

As is often the case, Jews find themselves caught in the middle. The future of antisemitism will not be decided on the streets, in synagogues, in Israel’s Ministry of the Diaspora, or even in the deliberations of international organizations such as OSCE. It will be decided in the conference rooms where technology policy is being shaped, and in the regulatory frameworks currently being drafted.

So, what must we do?

1. Be mindful of the threat posed by companies that are controlled by dictatorships, such as TikTok and DeepSeek, and understand that they may seek to undermine democracy and the West from within. 
2. Understand that despite the immense potential inherent in AI, the ability to use it for manipulation, disinformation, and swaying public opinion makes it a double-edged sword. Thus, alongside the accelerated development of AI, we need global agreements on restrictions on its invasive misuse.
3. Continue to strengthen privacy laws and data protection laws, especially regarding sensitive personal data, because the link between data and racism and antisemitism has never been stronger.

The unprecedented wave of antisemitism sweeping across social media, accelerated by AI, should serve as a warning sign of the potential catastrophe that lies ahead not only for Jews, but for all minorities and marginalized populations around the world. If protections are not implemented on a global level, that canary in the coalmine will not be the only one that suffocates.