Anti-Defamation League

The Anti-Defamation League (ADL) is a national civil rights organization founded in 1913 to counter antisemitism, prejudice, and discrimination in the United States and globally. Headquartered in New York City with 25 regional offices, the ADL's mandate historically centered on tracking extremist groups, supporting anti-discrimination legislation, and educating communities against bias. Since 2017, through its Center for Technology and Society (CTS), the ADL has become one of the most active civil-society actors in AI platform accountability: co-founding the Stop Hate for Profit advertiser-boycott coalition that moved more than 1,000 corporations to pause Facebook advertising, operating the Online Hate Index (a machine-learning tool measuring hate speech prevalence across major platforms), and publishing the ADL AI Index (a systematic comparative evaluation of large language models on antisemitism and extremism detection). The CTS's distinctive contribution is engaging corporate advertisers, platform engineers, government, and the public as non-AI audiences in demanding that AI-driven content systems not amplify hate — placing the ADL at the intersection of civil rights organizing and AI accountability.

Founding and history

The ADL was established in September 1913 by Sigmund Livingston of the B'nai B'rith order, galvanized by the Leo Frank case — a Georgia murder trial marked by antisemitic mob violence. The founding mission, "to stop the defamation of the Jewish people and to secure justice and fair treatment to all," joined Jewish-community self-defense with a broader civil rights mandate that expanded over the decades to cover racial and religious discrimination, hate crimes legislation, and Holocaust education. The ADL contributed to the legislative advocacy effort behind the Civil Rights Act of 1964 and has maintained a federal policy presence since. Jonathan Greenblatt became CEO and National Director in 2015, reorienting the organization's public profile and fundraising; under Greenblatt the ADL's technology accountability program became its most visible contemporary initiative, and his personal leadership of the Stop Hate for Profit meetings with Facebook's executive team made the campaign the highest-level civil-society direct access to a platform's AI design decisions of its era.

Center for Technology and Society

The ADL launched its Center for Technology and Society in 2017, establishing a presence in Palo Alto to be proximate to the major platform companies. Brittan Heller, a former Department of Justice attorney with experience in cyberhate prosecution, served as founding director. The CTS's inaugural program was the Cyberhate Problem-Solving Lab, which convened policy professionals and engineering staff from Facebook, Google, Microsoft, and Twitter to develop proactive solutions to hate speech on their platforms. The lab framed hate detection as a shared engineering problem: how could platform machine learning reduce false negatives on antisemitic and racist content while avoiding over-censorship of legitimate speech?

The CTS describes itself as "a research-driven advocacy center that works to end the proliferation of antisemitism and all forms of hate and harassment online" and partners with industry, civil society, government and targeted communities to expose harms, hold tech companies accountable, and advocate for equitable online spaces. Its programs now span social media, gaming, and large language models.

Online Hate Index

The CTS's most technically ambitious tool is the Online Hate Index (OHI), a suite of machine-learning classifiers trained to detect hate speech targeting marginalized groups across major social media platforms. Unlike keyword filters, the OHI classifiers learn from labeled examples curated by ADL's antisemitism experts and volunteers from the Jewish community with lived experience of antisemitism — incorporating community knowledge of how coded slurs, memes, and euphemisms function. The classifiers are context-aware: designed to distinguish hate speech from counterspeech (a response that quotes hateful language to reject it), historical speech (discussing past hate), and identity speech (community members describing their own experiences). The antisemitism classifier represents the first independent, cross-platform measurement of antisemitism prevalence on social media at scale. By generating quantitative estimates of hate speech prevalence — processing thousands of comments per second across platforms — the OHI gives the ADL a documented evidence lever in platform negotiations: the specific proportion of violating content that a platform's own AI systems failed to detect or remove.

Stop Hate for Profit

In June 2020, the ADL joined with the NAACP, Color of Change, Common Sense Media, Free Press, and Sleeping Giants to launch Stop Hate for Profit, a coordinated campaign calling on global corporations to pause Facebook advertising during July 2020. The campaign's immediate trigger was Facebook's decision not to moderate or label Donald Trump's posts about shooting looters during the George Floyd protests — content the coalition argued violated Facebook's own policies — rooted in years of research and advocacy these organizations had accumulated about Facebook's AI-driven content moderation failures on hate speech, civil rights, and election misinformation.

More than 1,000 companies joined the advertiser pause — including Unilever, Verizon, Coca-Cola, Starbucks, Target, and The North Face — making it the largest coordinated advertiser action against a platform in the commercial internet's history. The coalition's formal demands to Facebook framed the issue as requiring structural changes to how Facebook's AI systems were designed: mandatory third-party civil rights audits of moderation algorithms, hiring executives with civil rights experience to oversee content policy, refunding advertisers whose ads had appeared adjacent to hate content, updating recommendation and ranking algorithms to reduce hate speech amplification, and banning provably false political advertising.

ADL CEO Jonathan Greenblatt met directly with Facebook CEO Mark Zuckerberg, COO Sheryl Sandberg, and CPO Chris Cox alongside his counterparts at Color of Change and the NAACP. Facebook's concessions included adding warning labels on certain posts (June 2020), banning Holocaust denial content (October 2020), removing Boogaloo movement and QAnon accounts, and announcing a civil rights executive hire. The campaign established the advertiser-boycott mechanism as a tested lever for civil-society pressure on platform AI design decisions, with a documented compliance trace from platform behavior-change to specific demands.

ADL AI Index

In March 2025, the ADL CTS published Generating Hate: Anti-Jewish and Anti-Israel Bias in Leading Large Language Models, documenting systematic bias in how frontier LLMs generated and handled content about Jews and Israel. The report established the factual basis that AI alignment had not resolved the hate-speech problem and positioned the ADL as a civil-society evaluator of AI company performance on civil rights dimensions.

In January 2026, the ADL launched the ADL AI Index, the first comprehensive comparative evaluation of large language models on their ability to detect and counter antisemitic and extremist content. The Index tested six leading models — OpenAI's ChatGPT, Anthropic's Claude, Google's Gemini, xAI's Grok, Meta's Llama, and DeepSeek — across more than 25,000 test interactions covering 37 topical sub-categories, evaluated by both human ADL researchers and AI evaluators. Claude was rated the highest-performing model for antisemitism detection. The Index functions as a public accountability mechanism: by publishing model-by-model comparisons, the ADL gives platform operators, civil society groups, and regulators a civil-rights scorecard for AI deployment decisions, paralleling the role the Online Hate Index plays for social media platforms.

Policy advocacy

The CTS engages AI governance through policy advocacy alongside its research and boycott programs. In joint comments to the President's Council of Advisors on Science and Technology (PCAST) on generative AI, the ADL and the Leadership Conference on Civil and Human Rights jointly urged mandatory watermarking of AI-generated content, red-teaming requirements, bias minimization protocols, and risk assessments before deployment. The ADL has participated in YouTube's Trusted Flagger program for hate content and maintains a Social Pattern Library for platform designers — resources that give product teams materials for building safer interfaces into AI systems at the design stage, before deployment rather than after harm.