Palestinian human rights and civil society organizations have raised concerns over the dehumanization of Palestinians on Meta's digital platforms. The issue, exacerbated during the recent Israeli-Palestinian conflict, revolves around biased results generated by Meta's artificial intelligence tools. This dehumanization stems from bias in training datasets and reflects a broader problem of AI-driven discrimination against Palestinians.
Incidents such as WhatsApp's AI image generator creating gun-wielding children emojis in response to "Palestinian" and Instagram's AI-translation model replacing "Palestinian الحمد الله" with "Palestinian Terrorist" highlight the persistent issue. These results perpetuate negative stereotypes and contribute to the marginalization of Palestinian voices.
Palestinian content remains disproportionately over-moderated on Meta's platforms, stifling freedom of expression and restricting access to information, especially during crises. Recent changes to Meta's automated systems lowered the threshold required for action on Arabic/Palestinian content from 80% to only 25%. This has led to a surge in censorship and diminished the reach of Palestinian journalists and human rights defenders.
Furthermore, hate speech, incitement, and disinformation continue to spread across Meta's platforms, undermining the online safety and dignity of Palestinian voices. Meta's internal documents acknowledge that Hebrew hostile speech classifiers have been ineffective due to insufficient data.
Advocates for Palestinian digital rights demand immediate action from Meta to rectify these issues. They call for a comprehensive audit of the datasets used for AI training in content moderation to address concerns related to negative stereotyping. Transparency on criteria and thresholds for moderation, especially related to "Dangerous Organizations and Individuals," is essential to prevent over-moderation. Additionally, the activation of Hebrew hostile and hate speech classifiers is necessary to curb the spread of hate speech and incitement.
In conclusion, Palestinian rights advocates urge Meta to take immediate and decisive action to eliminate bias and discriminatory practices on its platforms. This is not only essential to uphold human rights online but also to regain the trust of Palestinian users and maintain credibility worldwide. The call is clear: Meta must do more to protect Palestinian digital rights and ensure a fair and safe online environment for all.