Strengthening Arab fact checking: lessons from partnering with the Arab Fact-Checking Network

As the conflict in the Middle East escalates dramatically, accurate information is at a premium. The internet is already awash with text and video that is either misleading or completely false, sometimes creating dangerous narratives that raise tension and cause harm.

Across the Middle East and North Africa, journalists and fact checkers have always operated in complex and often high-risk information environments. But at times like this, when conflict is spreading and political instability is heightened, there is fertile ground for false and misleading claims to spread quickly, often across borders and platforms.

For fact checking organisations, the challenge is not only identifying misinformation, but doing so fast enough to limit its reach and impact. That’s where Full Fact’s AI tools, which have been expanded successfully into Arabic, can help.

Full Fact worked for several years with the Arab Fact-Checking Network (AFCN) led by Arab Reporters for Investigative Journalism (ARIJ), to explore how AI-powered fact checking tools could support their work at scale, and to assess the impact they could have when adapted for a new language and regional context.

The AFCN now includes more than 35 member organisations working across the Arab world. Collectively, its members reach audiences exceeding 6 million followers across the Middle East and North Africa, giving them significant reach and influence in national and regional conversations.

A partnership grounded in regional realities

The partnership with ARIJ’s AFCN between 2023 and 2025 focused on piloting Full Fact’s AI tools in Arabic and understanding how they could be used by fact checkers working in diverse political and media environments across the region. The aim was not simply to deploy technology, but to test whether it meaningfully improved workflows, reduced time pressure and enabled new forms of fact checking.

In total, 25+ AFCN member organisations used Full Fact AI tools during the pilot period. Fifteen organisations completed pre- and post-use surveys, which form the basis of the findings described below.

These results reflect the period when the tools were available and used actively in 2024. Due to funding cuts from Google in 2025, AFCN members lost access to the tools just when they needed them most. Many organisations have highlighted how this has had a negative impact on their fact checking work.

Faster identification of check-worthy claims

One of the clearest findings from the evaluation was the effect the tools had on speed and efficiency.

Thirteen of the fifteen survey respondents said that Full Fact AI tools significantly accelerated their work and made it easier to identify claims that required fact checking. In practice, this meant less time spent manually monitoring content and more time available for verification and analysis.

Monitoring capacity also increased. Thirteen organisations reported that they were able to monitor a greater number of media outlets after using the tools, expanding their coverage without increasing staff time.

Unlocking live fact-checking

The ability to conduct live fact checking emerged as a major shift.

Twelve of the fifteen responding organisations said they could now fact check speeches and statements in real time and publish results shortly afterwards. Seven of these organisations reported that they were not able to do live fact checking at all before using the tools, but could now do so easily, saving significant time and effort.

This capability is particularly important in contexts where misleading claims are often made during long speeches or broadcast interviews, and where delays can allow false information to spread widely before it is challenged.

Significant reductions in monitoring time

Survey results also show a clear reduction in the time required to extract claims from interviews and speeches.

Before using the tools, only one organisation reported that media monitoring typically took less than an hour per day. After using Full Fact AI tools, eight organisations said they could extract claims in under an hour, representing a substantial improvement in efficiency.

For five organisations, Full Fact AI became their primary tool for monitoring and claim identification.

AI supporting, not replacing, human expertise

As with all Full Fact AI deployments, these tools are designed to support, not replace, human judgement. Fact checkers remain responsible for reviewing outputs, verifying claims, and providing the context and explanation that audiences rely on.

This balance is especially important in Arabic-language fact-checking, where dialects, political nuance, and cultural context play a critical role in understanding what a claim means and how it should be assessed.

By reducing the time spent on monitoring and transcription, the tools allow fact-checkers to focus their expertise where it matters most.

This is visible in the more than 145 fact checks which were published supported by Full Fact AI tools over a four month period. These covered a wide range of topics, including politics, economics, social issues, crises, and gender-related claims.

Adapting AI tools for Arabic

A significant part of the partnership focused on adapting the tools for Arabic-language content. This included annotation and model evaluation work to ensure claims were accurately identified across Modern Standard Arabic and regional dialects.

Fact checkers contributed to reviewing outputs and improving performance, helping refine how the system handled linguistic nuance, political context and culturally specific references.

This collaborative annotation process was essential to ensuring that the technology reflected the realities of Arabic media environments rather than simply translating an English-language system.

A model for regional and language expansion

The AFCN partnership provides a clear example of how Full Fact supports fact checking communities as it expands into new regions and languages. Rather than a one-size-fits-all approach, the work emphasised collaboration, training, and continuous feedback, ensuring the tools responded to real needs on the ground.

As misinformation increasingly crosses borders and platforms, supporting fact checkers in diverse linguistic and regional contexts is essential. The experience of working with the AFCN shows that when AI tools are thoughtfully adapted and paired with local expertise, they can play a meaningful role in strengthening the global fact checking ecosystem.

It also highlights the importance of sustainable funding models for fact checking infrastructure. When tools are withdrawn due to funding constraints, capacity is affected. Long-term impact requires long-term support.

As Rawan Damen, ARIJ Director General, put it: "The integration of advanced verification tools was a transformative milestone for fact-checkers across the MENA region, enabling us to combat disinformation in real-time. The current funding shortfall and no access to the tools has unfortunately reversed this progress, reverting our specialists to labor-intensive manual workflows. As information disorder reaches a fever pitch amidst regional uncertainty, we must find a way to bridge this resource gap to ensure our work remains both timely and impactful."

Full Fact continues to look for new funding to enable the tools to be made available again across the Arab world. As a charity, we are dependent on the support of partners who understand why this work is more important than ever.