Somewhere in the outskirts of London, an AI-based startup aimed at detecting bias against Israel in public media outlets worldwide, is being established. Why public media? Because it is obligated by law to provide balanced and fair reporting to taxpayers who fund it. Haran Shani-Narkiss, a brain researcher at UCL, gained attention about six months ago with the publication of a comprehensive report regarding the distorted coverage of the Israel-Hamas war in BBC broadcasts, and especially on BBC Arabic. But the dramatic report that was covered in the Daily Mail, The Telegraph, New York Post, Fox News, and others was just the opening shot.
Haran, let's start at the beginning: tell me about the group you established.
It started with a conversation with a friend who is also a brain researcher, like me, right after October 7th. We wondered if we could examine BBC coverage with real research. It's the largest media outlet in the world - a billion website visits every month - but it's also legally bound by a royal charter to be unbiased. To us - it's very clear that BBC is heavily biased against Israel, but pro-Palestinians claim that the BBC is actually biased against them.
Well, the pro-Palestinian narrative defines fairness as coverage that is 100% against Israel.
I think there's a bit more to it than that. We live in an era of post-truth. Within this framework, people say that the BBC is pro-Israel, because they can claim anything. So why not? The BBC was never called pro-Israeli – this claim is new. But as researchers, our goal is to discover the truth. So we established a research team to check if the BBC fulfills its commitment to cover the war in an unbiased manner. It's a complex problem - we needed to figure out how to quantify bias and what to compare it to. So we gathered academics and started to sketch a plan.
Were they all Israelis?
In the first stage - yes, and from across the whole political spectrum, including some very critical of the Israeli government, because it was important for us to understand how to neutralize our own bias. This is what scientists do - they understand what their bias is and figure out how to design the experiment so it won't be affected by it.
But there is so much bias in academia as well.
There is bias everywhere. Every research has biases, and every researcher has assumptions, and therefore, good research is one that overcomes bias. I'll give you an example - we had a feeling that the reporting often used vague language. For instance, if someone fired at Israel, they reported objects or projectiles flying in the sky, but if Israel fired, they reported very clearly on who fired what, and how many Palestinians were injured. The question was how to examine this in large numbers over time.
If the error is truly innocent, over time it should happen on both sides. So we conducted an experiment. We gave an article to an AI model, and it checked it according to questions we presented regarding how many Israelis were injured and how many Palestinians, and we did this with 1,500 articles. Alongside the model, we also had humans read the articles and examine whether they understood from the reporting whether there were Israeli casualties and whether there were Palestinian casualties. If the article didn't convey this clearly, the humans were asked to read more articles from the site from the same day and understand which casualties were involved. After we had all the reports checked by the model and by humans, we compared them. In each pairing, there could be overestimation, underestimation, or balance. When we counted this across 1,500 articles, we discovered that in over 70% of the cases, there was agreement between the AI model and what people understood. In terms of the type of errors, there was perfect balance when it came to Palestinians, compared to a chance ten times higher of getting an underestimation when it came to reporting on Israelis. This gives a good example of the meaning of quantitative research.
What do you think causes the bias?
I'm not even sure it's done consciously. I think there are some people who have such a deep bias that they simply are incapable of presenting Israelis as victims. But that’s irrelevant for the research. If you create a good index, you can design an experiment that isn't affected by your bias, and where the subjects can't cheat, because you ask them many questions, which leaves them not entirely clear as to what you're trying to find. This worked similarly in our experiment regarding headlines that express sympathy. We found that a person's opinion doesn't affect their answer to the question of whether a headline generates sympathy. If a headline says that Palestinian mothers are mourning the loss of their children, it doesn't matter who you support; this headline generates sympathy.
That's one example - but how do you build a model that deeply examines coverage over time? It's endless.
This is indeed just one example of what we wanted to create. We gathered scientists from the worlds of Large Language Models, and we wanted to find ways that verify empirically and scientifically across many parameters and using various methods. We started with a group of ten people from Israel, England, France, and the US. We met on Zoom and realized that this is actually a legal matter, so we started looking for lawyers who could take this to court in the UK. That's how we reached attorney Trevor Asserson, and decided to join forces - he would receive scientific advice from us and the AI and language tools, and we would receive legal guidance. We established RIME: Research for Impartial Media, and started weekly meetings. The peak came in February '24, when we held a hackathon - an event where 40 people from various fields come together to try to solve a problem in one concentrated day. Scientists came from Hebrew University, Ben Gurion University, Tel Aviv University, as well as Columbia, UCL, and Cambridge. We gathered researchers from computer science, analysts, intelligence experts, Arabic specialists, and experts in quantitative research around human language. We worked intensively for nine months, all voluntarily, and in September, the report came out. We achieved our goal.
What was your motivation?
Zionism. Justice. The desire to be able to sleep well at night, knowing that you gave your skills for what is currently happening in your country. People were looking for a way to contribute and channel the fear and alienation we felt at the university. On October 8th, I spoke with a colleague whom I really like, and he said some really harsh things. This was one day after the massacre, when I knew that I too could have been a victim there, too. The conversation was shocking. I literally felt my Jewish soul yearning. I knew this was my place and that this is what I needed to do now.
How long did you work together?
About nine months, all voluntarily. We dedicated ourselves completely, at the expense of other work and time with our families, out of an altruistic and Zionist desire. We weren't paid, but for Israelis in exile, like me, the feeling that we needed to do something was very strong.
Where does it stand now? Have you filed a lawsuit?
We've submitted the report, and the BBC has submitted a response saying it's nonsense. We've appealed this response, and we're waiting for a response to our appeal. It's going to be a long and cumbersome process. There won't be a moment when everything changes at once. But there might be a moment when the UK government will understand that regulation is needed for the public broadcast in different languages in England, in which only God knows what's happening (see box).
When did you decide to apply this model to other media outlets as well?
At some point, I made a pivot in my life path. I was a brain researcher, and today I'm essentially developing methods based on large language models to quantify variables from large amounts of text. I've been working with AI models for quite a few years now, and for me, the new language models are the way to understand what's happening under the hood in today's media.
So where are you going with this, essentially?
I'm using these tools for research on other public broadcasting corporations around the world, to quantify bias in places where Jewish communities feel it exists, and works against them. Canada, Australia, Belgium, and more. I'm essentially saying - enough of complaining "the Norwegians hate us the most," or "the Spanish are the most antisemitic." We need to know what is really being said about us, at least in public broadcasting corporations worldwide, because in public broadcasting, there's a charter and there's an obligation for fair reporting, so legally, one can act if bias can be proven.
Why do you think there isn't a coordinating body doing this in an organized way for all Jews, or all Israelis?
Excellent question. I don't know.
Do you think the Foreign Ministry should take this on as part of its new PR budget?
Seems very logical to me. It's the best tool that exists today, which can refine needs and direct the government on how to act in the PR aspect. It provides tools to attack real, empirical, quantitative problems.
If the state were to take this on - what kind of budget are we talking about?
The question is what they would want to check. If the Foreign Ministry really wanted to check large public media outlets in key countries, my estimate is that something in the area of a million dollars would be enough. I was happy to hear that a large budget was recently announced to improve Israel's image abroad. I believe that such a project at the national level could be critical and could channel the big money much more efficiently.
And where does the company you established stand today?
I'm currently going along two tracks - in the personal track, I'm in the process of leaving academia and establishing the new company I founded, InnoHives. The second track is voluntary and related to the research we did - I lecture and promote the topic, together with Trevor Asserson, and we continue the journey with a more limited group facing organizations and lobbyists, to spread what we discovered about the BBC. Jews are not the only community suffering from this.
Who else suffers?
For example, the Hindu community. They also feel that in everything related to Muslim-Hindu tensions in India, the BBC covers it in a very unfair way. One side is always presented as the victim and the other side as the aggressor, while the situation is far from it. In general, there's a feeling that the BBC holds very narrow agendas and a very dictated perspective. That's fine as a private media outlet, but not as a public media outlet that pretends to report without bias.
Do you feel that the BBC report had an impact?
Recently, when the story about the documentary on BBC based on the son of a Hamas senior official exploded, I understood that the report has a long life, because people turn to it to validate what they see and feel, including members of Parliament. They still contact us, requesting briefings and references. The report enabled people to receive validation, even the very understanding that they're not crazy, that there really is something messed up here, and it's not just their imagination.
One of the safe grounds a person wants to stand on is truth. But today we don't have the truth. In such a situation, when your neighbors say 'Israel is like Nazi Germany,' you no longer know if you're crazy or sane. In order to act correctly, we simply need this tool.
"The British taxpayer cannot fund the prevalent antisemitism in Arab countries"
One of the main contributors to the research on Arabic coverage is national security and foreign policy researcher Khaled Hassan. An Egyptian-British convert to Judaism who is now an expert on digital dangers, radicalization, disinformation, and antisemitism in the Arab world.
"I regularly research terror support and digital threats in the UK," he says, "so when a friend introduced me to Trevor Asserson, and through him to Haran, the connection was natural."
Hassan began following the BBC Arabic channel back in 2019, after the broadcast of a documentary called "One Day in Gaza," in which one of the subjects referred to certain music that 'makes you want to slit a Jew's throat,' and in the translation the word 'Israeli' appeared instead of 'Jew.' "BBC in Arabic is essentially a British Al Jazeera," explains Hassan in a restrained British accent, "it's a pro-terrorism channel that is anti-Israel."
Hassan is very concerned about the channel, which, according to the report, 96% of its content was biased in favor of Palestinian narratives. "The channel has a large viewership and enormous influence - it is quoted as a reliable source by researchers and media professionals in Arab countries, and in fact all over the world. I wrote to the BBC's General Secretary when I discovered the misleading translation, and they insisted on not correcting it, claiming that the translation refers to the subject's intention and not to the literal translation. This is one small example, but the problem is that over time, it affects policy. Arab politicians demonize Israel based on what is aired on BBC Arabic, while its image is of a moderate and accurate media outlet, and that's what's so dangerous."
Hassan's goal is for the Conservative Party, which he advises, to return to power in four years and stop public funding for the BBC Arabic channel. "It is unacceptable for the British taxpayer to fund antisemitic sentiments that exist in countries like Egypt or Jordan. This also preserves opposition to normalization. In Saudi Arabia and the Emirates, for example, the media outlets are more balanced and try not to talk about Israel in the way BBC Arabic presents things."