Engineering Disinformation: Women in AI’s Crosshairs

By Nahila Salameh

Within minutes, fabricated images were enough to set social media ablaze, turning online spaces into campaigns of accusations and speculation.

On April 11, 2026, an image circulated showing Lebanon’s ambassador to Washington, Nada Mouawad, inside Ben Gurion Airport in the Israeli entity. She appeared to be wearing a scarf combining the Lebanese cedar with the Star of David, in an apparent attempt to suggest that she had visited the occupied territories. However, Fact Check Lebanon, a platform affiliated with Lebanon’s Ministry of Information, revealed that the image had been generated using artificial intelligence, citing clear signs of manipulation, most notably distortions in the background and inconsistencies in lighting and shadows.

This was not the only campaign targeting Ambassador Nada Mouawad. On April 15, 2026, another image circulated, claiming to show her shaking hands with the Israeli ambassador to the United States following a meeting in Washington, with the Lebanese and Israeli flags visible in the background. Fact Check Lebanon, affiliated with the Lebanese Ministry of Information, again confirmed that the image was false and had been generated using artificial intelligence.

Ambassador Mouawad also confirmed in a press statement that the meeting “did not involve a handshake” with her Israeli counterpart.

Nada Mouawad’s case places us before a new phase of gender-based digital violence, in which artificial intelligence tools are being deployed in multiple ways to influence women’s visibility in public space and their participation in decision-making.

The roots of these campaigns, however, date back to 2019. With the outbreak of popular protests in the Lebanese streets, women emerged at the forefront, and campaigns of defamation and digital violence against women journalists, activists, and politicians began to intensify. These attacks sought to silence their voices and reframe their presence within the boundaries of what was deemed “permitted” and “forbidden,” contributing to a shrinking space for public debate and greater restrictions on freedom of expression.

Since then, social media platforms have witnessed a marked increase in this form of violence, fueled by fabricated news, hate speech, manipulated images, and threats of murder and rape.

Generative artificial intelligence has added a new dimension to an already existing pattern. These tools are now capable of producing videos, images, and audio recordings that are difficult to distinguish from authentic material. When such content is recirculated without verification, the impact of these campaigns multiplies, further intensifying the targeting of women with the aim of excluding them from public space.

Artificial intelligence tools are also being used in multiple ways. Their role extends from generating fabricated content itself to the algorithms that govern its circulation. Social media platforms rely on AI-based recommendation systems designed primarily to maximize the amount of time users spend engaging with content, rather than to assess its accuracy or credibility.

Because sensational, shocking, or polarizing content generally generates higher levels of engagement than balanced material, these algorithms tend, without any direct editorial intent, to amplify fabricated images and false claims more quickly and widely than the fact-checks that later debunk them.

In this way, artificial intelligence plays a dual role in engineering such campaigns: it serves as a tool for producing content on the one hand, and as a tool for determining who sees that content and when it spreads on the other.

These campaigns further evolved during Israel’s war on Lebanon between 2024 and 2026, which became fertile ground for the spread of hate speech and misleading content.

So what does it mean to engineer a campaign? Why are women targeted by hostile campaigns that can go so far as to destroy their reputations and personal lives? And how does artificial intelligence contribute to the evolution of these campaigns, expanding their reach and magnifying their impact?

Women as Targets of Disinformation

This phenomenon intersects with the findings of a recent study by the Samir Kassir Foundation’s SKeyes Center for Media and Cultural Freedom, titled The Price of Freedom: Gendered Digital Harassment and Democracy in Lebanon (January 2026). Based on nine in-depth interviews with Lebanese women journalists and politicians, the study found that seven of the nine participants reported that the effects of digital campaigns had spilled over into their offline lives, while more than half had received direct death threats.

The study indicates that these coordinated campaigns are fueled by digital networks affiliated with multiple political parties, as well as media outlets close to banking circles. This challenges attempts to attribute the phenomenon to any single political actor and instead points to a structural pattern that cuts across partisan affiliations.

With regard specifically to artificial intelligence, the study describes its current role as “a new layer beginning to emerge within this ecosystem, but one that has not yet become dominant.” Participants reported exposure to manipulated screenshots, fabricated images, and suspected automated amplification of content, without detailed documentation of cases involving content generated entirely by AI. The study warns that the spread of generative AI tools could eventually turn Lebanon into “a testing ground for AI-generated gendered disinformation.”

Against this backdrop, the case of Ambassador Nada Mouawad can be seen as an early warning of how far smear and disinformation campaigns targeting women in Lebanon could go when AI tools are incorporated into an already established ecosystem of gender-based digital violence. The following four cases illustrate manifestations of precisely this ecosystem, the fertile ground SKeyes warns about, regardless of the tools used to produce the content, and without evidence in any of these cases of AI-generated material specifically.

In June 2026, MP Paula Yacoubian sparked widespread controversy after signing a “Call to Save Lebanon,” which urged support for negotiations to end the war, bring the Israeli occupation to an end, restore the Lebanese state’s sovereign decision-making power, and ensure that weapons remain exclusively under state authority. Following reports about the appeal, her statements circulated widely across X, Facebook, and Instagram, where she was subjected to campaigns of accusation, defamation, and hate speech. She was accused of normalization and of abandoning the “resistance,” even though the appeal she signed also called for an end to the Israeli occupation.

On March 5, 2026, An-Nahar also documented what appeared to be an organized campaign against Yacoubian. The campaign accused her of opposing Hezbollah while allegedly hosting Iranian nationals at a hotel owned by her friends, without presenting clear evidence. The hotel’s owners denied that any non-Lebanese guests were staying there.

Journalist May Chidiac was likewise subjected to a violent online campaign in September 2025, including insults, intimidation, and threats, after criticizing positions linked to Hezbollah. According to a report by L’Orient-Le Jour, the campaign spread across social media platforms and directly targeted Chidiac. A video posted on X contained severe personal insults and mocked her disability. It garnered hundreds of thousands of views, further expanding the reach of the attack. The case offers a stark example of how targeting can shift from a woman’s political position to her body and disability.

MP Halima El Kaakour was similarly targeted ahead of the January 8, 2025 parliamentary session to elect a president. Speaking on Al Jadeed, she announced that she would vote for Ziad Baroud and asked directly: “How can a president swear an oath to uphold the Constitution when he comes to office through a violation of the Constitution?”—an implicit reference to the candidacy of army commander Joseph Aoun, whose election would have required a constitutional amendment.

Her statement immediately circulated on X, prompting singer Elissa to respond with a sarcastic post: “Unbelievable how she became an MP!” The comment triggered a broad wave of hostile responses against El Kaakour from Elissa’s followers.

Journalist Sahar Ghaddar also became the target of such campaigns. In mid-December 2024, during Israel’s war on Lebanon, Ghaddar, together with activist Ghina Ghandour, posted on X accusing a Lebanese television station of displaying media bias in favor of “Israel” in its coverage of the war. The two women refused to delete their posts, after which the station filed a defamation lawsuit against them and they were summoned by Lebanon’s Cybercrime and Intellectual Property Bureau. News of the case then spread amid a counter-campaign by supporters of the station, who accused the activists of incitement.

 Digital Security Beyond Control

What can be established is that artificial intelligence tools have opened up a new frontier for digital campaigns, enabling the production of more convincing disinformation, accelerating its circulation, and expanding its reach, while platform algorithms contribute to amplifying inflammatory content. For women, the consequences are both psychological and professional, potentially pushing them to scale back their presence in public spaces or self-censor what they post. This is precisely what Zeinab Ismail, Senior Policy Analyst and Project Manager at SMEX, warns against.

“The danger of artificial intelligence lies in its ability to produce fake images and videos, and in the fact that many users are unable to distinguish authentic content from AI-generated material,” she says. “This makes these tools an effective means of defaming women, undermining their credibility, silencing their voices, and sending an intimidating message to any woman considering participation in public life. These campaigns are all the more dangerous because content posted online can continue to circulate even after it has been deleted, prolonging the harm inflicted on victims.”

Zeinab stresses that “confronting this form of disinformation begins with verification before publication and reliance on fact-checking platforms, alongside training journalists in digital verification skills and in detecting AI-generated content. It is also important to report any misleading or fabricated material, avoid engaging with blackmail or defamation campaigns, document violations, and seek assistance from relevant authorities or digital support organizations.”

She adds that “Lebanon’s legal framework is still unable to keep pace with this type of violation. Existing laws are outdated and fail to address the challenges posed by digital platforms and artificial intelligence, while the National Cybersecurity Strategy has yet to be implemented as required.”

She also points out that “social media platforms themselves prioritize engagement and reach, which can contribute to amplifying misleading and abusive content targeting women. At the same time, their algorithms may restrict or remove credible content that does not conform to their standards.”

Yet while the mechanics of campaigns waged across social media platforms can be explained technologically, and while the absence of effective deterrent legislation allows them to flourish, it is their gendered dimension that explains both their intensity and the way they are directed. Farah Salhab, an independent consultant specializing in gender and inclusion, explains why.

Gender-Based Violence in the Digital Sphere

Farah argues that “discrimination against women in digital spaces cannot be separated from discrimination in public life. Stereotypes that cast doubt on women’s fitness to lead or appear publicly are carried over into digital platforms, where they are amplified and can reach vast audiences within a very short time.”

She explains that “the targeting of women differs from the attacks men may face. It does not stop at criticizing their positions or opinions, but extends to undermining their very right to occupy public space. Their appearance and private lives are targeted, while sexualized insults and real or fabricated images are used to discredit them and transform them from women with positions and opinions into objects of defamation. These campaigns seek to reimpose traditional gender roles, pushing women into silence, reducing their visibility, or driving them out of media, politics, and public life altogether.”

She points to a 2025 UN Women survey showing that “69.9% of the women surveyed experienced digital violence in the course of their work, rising to 74.9% among women journalists and media professionals.” The survey also found that “40.9% of respondents experienced offline harm that began with digital violence, a figure that rose to 42% among women journalists.” In politics, an Inter-Parliamentary Union study on the Asia-Pacific region, published in March 2025, found that “60% of the women parliamentarians surveyed had experienced gender-based online attacks, while 44% reported receiving threats of murder, rape, physical assault, or abduction.”

The research further underscores that “artificial intelligence did not create violence against women, but it has transformed its tools and accelerated its spread.” Fabricated images, videos, and audio recordings can now be produced with ease, identities can be impersonated, and automated accounts can be used to recirculate and amplify abusive content. According to the UN Women survey, “23.8% of respondents said the violence they experienced involved the use of artificial intelligence technologies,” while a 2026 UN Women report found that “6% of women had been subjected to fabricated or manipulated images or videos produced using these technologies.”

Farah Salhab concludes that “one of the most dangerous consequences of these campaigns is that they push women toward self-censorship and curtail their participation in public life.”

This is where fact-checking platforms assume a crucial role: exposing manipulation in published content and reinforcing the importance of verification before material is published or shared. Youssef Al-Amin, Managing Editor of the Sawab fact-checking platform, explains how.

Debunking Claims and Unpacking Campaign Design

Youssef explains that “the role of fact-checking platforms extends to documenting the entire verification process, explaining the evidence used, and making sources available to the public, thereby strengthening transparency and trust in the verification process itself. Alongside this comes an educational dimension: improving media and digital literacy and explaining how disinformation works and the techniques it employs, so that individuals develop the critical-thinking skills needed to assess information before sharing or engaging with it.”

He argues that “the rapid evolution of generative artificial intelligence has made this role particularly important. Automated tools for detecting fabricated content still face limitations in accuracy and reliability, especially as generative models continue to advance. This makes the role of the human fact-checker all the more crucial, as it combines technical tools with contextual analysis, source verification, and an understanding of the political and social environment in which content is produced. The challenge has therefore become as much editorial and methodological as it is technological.”

From his position as a fact-checker, Youssef recommends adopting a simple principle: “Pause, then verify” before sharing any content. Information that aligns with a user’s beliefs or provokes a strong emotional response is not necessarily true, and cognitive biases can lead anyone, often unintentionally, to contribute to the amplification of a disinformation campaign. He specifically recommends tracing content back to its original source, analyzing its context, seeking confirmation from independent and credible sources, and avoiding reliance solely on platform algorithms or engagement metrics as indicators of credibility.

As for how organized disinformation campaigns operate, Youssef explains, based on Sawab’s monitoring, that they are typically engineered through six successive stages: “defining the objective—whether to damage a reputation, influence public opinion, provoke division, or undermine trust in an institution; analyzing the target audience by studying its interests, biases, and preferred platforms; producing the content, whether through misleading information, AI-generated material, or authentic content stripped of its original context; distribution and amplification through networks of real or fake accounts, coordinated groups, and paid advertising; recirculation by republishing the material in different formats or translations to extend its lifespan; and finally, measuring impact and adapting the campaign by monitoring reach and adjusting the narrative when it fails to achieve its objectives.”

He concludes that “understanding the architecture of a campaign as a step-by-step process helps curb its impact proactively, rather than merely correcting false information after it has already spread.” From his perspective, countering disinformation cannot be achieved by correcting false content alone, but rather “by building an audience equipped with the tools to evaluate information and make decisions based on knowledge and evidence.”

Gender-Based Violence in the Law

In May 2024, the European Union adopted Directive (EU) 2024/1385, which explicitly criminalizes the non-consensual production or creation of fabricated sexual content, including material generated using artificial intelligence, alongside cyber harassment, cyberstalking, and gender-based hate speech. The directive requires EU member states to transpose these provisions into national law by June 2027.

Lebanon’s legal framework, however, continues to lag behind international legislation addressing digital violence, particularly as a proposed law on protecting women from digital violence has yet to be adopted. The bill was put forward by FEMALE, in cooperation with the United Nations Population Fund (UNFPA) and other international organizations, on February 25, 2026.

The proposal bears the signatures of MPs Bilal Abdallah, Paula Yacoubian, Jamil El Sayyed, Nada Boustani, Halima El Kaakour, Sajih Attieh, and Michel Douaihy. It seeks to criminalize five key practices that are not clearly addressed under Lebanese law: cyberstalking, online harassment, digital identity theft, online extortion, and the non-consensual dissemination of intimate images or videos, including fabricated or AI-generated content.

Caught between Lebanon’s failure to enact adequate domestic legislation and its failure to sign or effectively implement relevant international conventions, women remain trapped in a legal protection gap, one made all the more dangerous by the inability of existing frameworks to keep pace with the speed at which artificial intelligence tools are evolving and the risks they pose.

By: Nahila Salameh

Disclaimer:

“This report was produced with the support of the European Union and the Government of the Netherlands as part of the Artificial Intelligence in Media Fellowship for the Middle East and North Africa. Its content does not necessarily reflect the views of the European Union or the Government of the Netherlands.

Generative artificial intelligence tools were used in the preparation of this work. The content was reviewed in full, and the accuracy of every piece of information and every source cited was verified prior to publication. Every effort was also made to identify and remove potential bias from the output, whether in the language used, the framing of events, or the selection of sources. Full editorial responsibility for this work rests with its author.”