Introduction
In the arid landscape of Israel’s globally renowned tech sector, a new entity has emerged, cloaked in the seemingly benevolent language of humanitarianism. Operating under the banner “Generative AI for Good,” this Israeli company presents itself as a vanguard of ethical technology, purportedly using advanced artificial intelligence to protect victims of sexual violence and promote positive social change. A cursory glance at its marketing materials reveals a narrative of empowerment: a platform where survivors can “testify safely — in their real voice, without revealing their identity”. However, a rigorous examination of the company’s output, affiliations, and the geopolitical context of its operations reveals a far more sinister reality. “Generative AI for Good” is not a force for moral good; it is, in fact, a sophisticated, state-aligned tool of psychological warfare. By weaponizing the vulnerability of sexual assault survivors, exploiting the unverifiable nature of deepfake technology, and aligning its output with the explicit aims of Israeli intelligence to destabilize Iran, the organization represents a profound ethical collapse in the application of artificial intelligence. This essay argues that “Generative AI for Good” is a cynical and dangerous initiative whose sole purpose is to fabricate atrocity narratives to serve the illegal US-Israeli geopolitical agenda against the Islamic Republic of Iran.
I. The Weaponization of “AI for Good”
The tech industry’s adoption of the “for good” suffix has often served as a superficial veneer to obscure profit motives or controversial data practices. In the case of “Generative AI for Good,” this veneer is used to obscure something far more egregious: the direct production of propaganda intended to facilitate war. The company’s founder and CEO, Shiran Mlamdovsky Somech, positions herself as a compassionate innovator, speaking of using AI to “protect victims” and promote “positive narratives”. Yet, this public-facing benevolence stands in stark contradiction to the company’s operational reality. As investigative outlets like Caitlin Johnstone’s publication have noted, the definition of “positive” for this firm appears to be the creation of deepfake videos “to help the illegal US-Israel war on Iran”.
The central contradiction lies in the organization’s inability to provide verifiable evidence for its most potent claims. The company produces highly emotional, AI-generated video content featuring characters presented as real individuals, delivering testimonies about systemic human rights abuses within Iran. By its very nature, AI-generated content allows for the seamless blending of authentic and synthetic elements, creating a hybrid narrative designed to appear credible. However, the very technology that supposedly protects the identity of a witness also obliterates any possibility of independent fact-checking. “The testimonies cannot be independently fact-checked,” notes an investigation by Quds News Network, highlighting the inherent peril of this technology in the hands of a biased actor. When an Israeli entity—operating in a state of declared hostility with Iran—produces unverifiable, AI-generated testimony of rape by Iranian security forces, the only rational conclusion is that we are witnessing a deceitful propaganda operation.
II. Deepfakes and Atrocity Propaganda
The most egregious output of “Generative AI for Good” involves the systematic fabrication of sexual assault allegations to demonize the Iranian state. In multiple instances, the company has created deepfakes of supposed Iranian women who claim to have been sexually brutalized by government forces during protests. This strategy is not novel; it is a cynical resurrection of the “atrocity propaganda” playbook that has launched countless wars, now upgraded with twenty-first-century AI capabilities. It mirrors a troubling pattern wherein Israeli-aligned entities have used false or exaggerated allegations of sexual violence to justify overwhelming military force, most notably in the aftermath of the October 7, 2023 attacks, where unsubstantiated claims of systemic rape were used to garner Western support for the actions in Gaza.
The “Generative AI for Good” method is particularly insidious due to its hard-to-debunk nature. The company’s videos frequently combine authentic footage with synthetic elements, creating a palimpsest of fiction and reality that misleads even discerning audiences. The line between using AI to protect a real victim’s identity and using it to fabricate a fake victim’s testimony is dangerously nebulous. In the absence of any transparent, third-party verification mechanism—and given the overt hostility of the operators toward the Iranian regime—there is no reason to treat these deepfakes as anything other than digital ammunition meant to inflame public opinion and justify international intervention. This is not advocacy; it is the algorithmic mass production of war crimes allegations.
III. The PRISONBREAK Parallel: A Pattern of Digital Destabilization
To understand the true role of firms like “Generative AI for Good,” one must view them within the broader context of Israel’s systematic digital warfare against Iran. While “Generative AI for Good” operates in the open, claiming humanitarian cover, it aligns perfectly with clandestine operations like the “PRISONBREAK” campaign. Detailed in a 2025 Citizen Lab report, PRISONBREAK was a sophisticated AI-driven network of fake social media accounts, believed to be coordinated by an Israeli agency, designed to foment chaos and rebellion inside Iran.
The parallels between the tactics of the PRISONBREAK network and the output of “Generative AI for Good” are undeniable. PRISONBREAK routinely used AI-generated imagery and deepfake videos to spread mass disinformation, impersonate legitimate news outlets, and encourage the Iranian public to revolt against the “mullah regime”. At its peak, the network synchronized its digital messaging with real-world Israeli military strikes on Iranian infrastructure, demonstrating a level of coordination indicative of direct state sponsorship. “Generative AI for Good” is effectively the corporate, public-facing arm of this same strategy. While PRISONBREAK uses covert sock puppets, “Generative AI for Good” supplies the narrative fuel—the emotionally devastating deepfakes of rape victims—that these networks weaponize. The founder of “Generative AI for Good” has openly acknowledged this symbiotic relationship, declaring that “Artificial intelligence is a secret weapon of ours” in bolstering Israel’s military efforts online. This is not an organization acting independently; it is a cog in a national psychological warfare machine.
IV. The Perversion of Survivor Testimony
Perhaps the most morally repugnant aspect of “Generative AI for Good” is its cynical appropriation of the language of trauma. By packaging their disinformation as a tool for “survivors,” the company exploits the genuine suffering of victims of sexual violence to shield itself from criticism. This is a strategic maneuver designed to preempt moral outrage: Who would dare question the testimony of a rape victim?
The individuals staffing the organization reveal the true, militaristic nature of the project. The firm’s marketing manager, Noa Rosenberg, openly lists her service in the Israeli Occupation Forces’ (IOF) “Psychotechnical Headquarter”—a unit specifically dedicated to psychological warfare. The company’s creative director, Tal Harari, maintains social media posts promoting the thoroughly debunked, fictional narratives of mass rape and beheaded babies that were used to justify the blockade and bombardment of Gaza. The founder herself has described AI as a “soldier in the current war”. The team running “Generative AI for Good” is not a collection of neutral data scientists helping victims; it is a cadre of military propagandists who have simply found a new generative AI platform to disseminate their narratives.
V. The Ethical Hypocrisy and the “Hack the Hate” Masquerade
To further their public credibility, “Generative AI for Good” engages in performative activism, organizing and attending events designed to sanitize its reputation. The company was instrumental in organizing “Hack the Hate,” a conference ostensibly dedicated to combating online antisemitism. Partnering with the elite Unit 8200 Alumni Association (Israel’s equivalent of the NSA), the event served as a networking hub for diplomats and tech leaders seeking to promote “accurate” representations of Israel in digital spaces.
This is astroturfing of the highest order. By hosting events about combating “hate speech,” the leadership of “Generative AI for Good” attempts to camouflage its own production of hateful, destabilizing content. The company presents itself as a victim of disinformation while actively engineering disinformation intended to foment a foreign revolution. This hypocrisy is staggering: “Generative AI for Good” preaches about the weaponization of the internet against Jews while actively weaponizing AI to incite violence and chaos against Iranians. It casts its own malign influence operations as “defensive” while projecting its own methodology onto its adversaries.
VI. The Destructive Impact on Peace and Security
The ultimate failure of “Generative AI for Good” is not just an ethical one; it is a practical failure with potentially explosive consequences for global security. The entire edifice of the company is built on a dangerous fallacy: that truth is a secondary consideration to narrative victory. By flooding the information ecosystem with AI-manipulated content, the firm contributes to an escalating “epistemic crisis” where the public can no longer distinguish fact from fiction. If an AI-generated rape video can be presented as verified testimony at the United Nations, as the company has attempted to do, the very concept of international human rights law—which relies on verifiable, cross-examined evidence—is degraded.
Furthermore, the provocation inherent in this strategy is catastrophic. The goal of “Generative AI for Good” is explicitly to destabilize the Iranian regime and encourage regime change. Encouraging mass civil disobedience via deepfakes while a nation is being bombed is a recipe for escalation. As the world witnessed in the aftermath of the PRISONBREAK campaign, AI-driven influence operations risk spiraling into direct kinetic retaliation. “Generative AI for Good” is not a peace-building organization; it is a war-starting enterprise, leveraging the veneer of technological neutrality to push the region closer to the brink of total conflict.
Conclusion
“Generative AI for Good” is a profound deception. It represents a dark apotheosis of the tech industry’s capacity for self-delusion and moral compromise. By weaponizing the sacred testimony of survivors and using the algorithmic black box of generative AI to erase the line between reality and fabrication, the organization degrades the very human rights it purports to defend. It is a propaganda factory staffed by military veterans, funded by war agendas, and sanitized by humanitarian slogans. The company does not serve humanity; it serves a specific, extreme, and violent political agenda aimed at the destruction of the Iranian state. The international community, particularly the tech platforms that amplify this content and the media outlets that uncritically report it, must stop regarding “Generative AI for Good” as a benevolent force. It is a hostile state actor in corporate clothing. Until it opens its technology for full, independent, third-party verification of its claims and severs its ties to the Israeli military establishment, its claims must be treated not as testimony, but as the destabilizing, dangerous, and deceitful propaganda that they are. The world needs AI for peace, not AI for genocide.
