{"query_echo":{"q":"","filters":{"analysis_type":["algorithmic audit"]},"warnings":[{"code":"facet_value_not_in_vocabulary","message":"Value 'algorithmic audit' for facet 'analysis_type' is not in the controlled vocabulary."}]},"total":{"value":3,"relation":"eq"},"page":1,"page_size":20,"self":"https://climateinsightsdirectory.org/api/v1/search?analysis_type=algorithmic+audit","stability":"beta","ai_marking":{"digital_source_type":{"uri":"http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia","name":"Created using Generative AI"},"notice":"Summary and key insights are generated using artificial intelligence."},"results":[{"id":"isdglo-000120","url":"https://climateinsightsdirectory.org/document/isdglo-000120","title":"Networks of Dissuasion","publisher":{"name":"Institute for Strategic Dialogue","organization":"institute-for-strategic-dialogue"},"document_type":"report","language":"en","date_published":"2024","source_url":"https://www.isdglobal.org/wp-content/uploads/2024/01/Networks-of-Dissuasion-Mapping-Online-Attacks-on-Reproductive-Rights-in-FR.pdf","generator":"Climate Insights Directory","generated_fields":["summary","key_insights","geography_discussed","analysis_type"],"summary":{"text":"The report analyzes anti-abortion activity across Facebook, Instagram, YouTube, and X in France from May 2022 to June 2023. It finds that anti-abortion mobilization relies heavily on 'dissuasive content'—including unverified testimonials and graphic imagery—and misinformation about abortion procedures. The research highlights a strong intersection between anti-abortion actors and far-right, anti-LGBTQ+, and religious communities. It specifically identifies the role of algorithmic recommendations on YouTube and Instagram Reels in serving harmful content to users, including minors, and documents the monetization of this content through Meta ads, with one page alone spending over €40,000 to reach millions of impressions, including targeting 13- to 17-year-olds with distressing imagery.","ai_generated":true,"unit":"isdglo-000120#summary","date_modified":"2026-10-23"},"key_insights":[{"text":"Anti-abortion mobilization in France primarily utilizes misinformation, misleading claims about abortion procedures, and 'dissuasive content' designed to discourage users from seeking care. This includes misrepresenting the suffering of the foetus, side effects for women, and the abortion process itself.","ai_generated":true,"unit":"isdglo-000120#insight-1","date_modified":"2026-10-23"},{"text":"Anti-abortion networks are closely interlinked with other ideological groups, with far-right accounts representing the largest community involved in amplifying this content. Other connected groups include anti-surrogacy, anti-LGBTQ+, and anti-trans advocacy organizations, as well as certain Christian and royalist actors.","ai_generated":true,"unit":"isdglo-000120#insight-2","date_modified":"2026-10-23"},{"text":"Meta generated an estimated €43,750 from 199 anti-abortion Facebook ads between May 2022 and June 2023, resulting in 9.4 million impressions. The page 'IVG Vous Hésitez' was the primary advertiser, spending an average of €40,033.","ai_generated":true,"unit":"isdglo-000120#insight-3","date_modified":"2026-10-23"},{"text":"Anti-abortion advertising on Facebook specifically targeted minors; 50 ads reached 13- to 17-year-olds, and 24 of those contained graphic or emotionally distressing content, including discussions of sexual assault.","ai_generated":true,"unit":"isdglo-000120#insight-4","date_modified":"2026-10-23"},{"text":"YouTube's recommendation algorithms served misinformative, graphic, and dissuasive content to all tested profiles, including a minor who had not previously watched anti-abortion content. The report also found that YouTube's promised 'informational panels' for abortion-related videos were rarely applied in practice.","ai_generated":true,"unit":"isdglo-000120#insight-5","date_modified":"2026-10-23"},{"text":"Instagram Reels recommended dissuasive or misleading content to users in one-fifth of the analyzed home feed posts. The algorithm also recommended 'tradwife' accounts that promote traditional gender roles and have links to far-right ideology.","ai_generated":true,"unit":"isdglo-000120#insight-6","date_modified":"2026-10-23"},{"text":"Anti-abortion actors on X (formerly Twitter) weaponized the debate over entrenching abortion rights in the French constitution and spread misinformation that mischaracterizes World Health Organization (WHO) guidelines.","ai_generated":true,"unit":"isdglo-000120#insight-7","date_modified":"2026-10-23"},{"text":"Some anti-abortion pages use deceptive tactics to appear neutral or pro-choice, such as using pro-choice hashtags and misleading titles, while actually promoting anti-abortion views. This may violate French law regarding 'obstruction to abortion'.","ai_generated":true,"unit":"isdglo-000120#insight-8","date_modified":"2026-10-23"},{"text":"The report identified the promotion of 'abortion pill reversal'—a scientifically unproven and harmful procedure—through Meta ads and YouTube videos, noting that platforms failed to remove this medical misinformation.","ai_generated":true,"unit":"isdglo-000120#insight-9","date_modified":"2026-10-23"}],"climate_tags":[],"geography_discussed":["United States","Poland","Hungary","Paris"],"frame_perspective":[],"analysis_type":["social media network analysis","algorithmic audit","ad library analysis","qualitative content analysis"]},{"id":"isdglo-000138","url":"https://climateinsightsdirectory.org/document/isdglo-000138","title":"Investigation 4 Spanish-Language News Consumers","publisher":{"name":"Institute for Strategic Dialogue","organization":"institute-for-strategic-dialogue"},"document_type":"report","language":"en","date_published":"2024","source_url":"https://www.isdglobal.org/wp-content/uploads/2024/06/Pulling-Back-the-Curtain-Spanish-Language-News-Consumers.pdf","generator":"Climate Insights Directory","generated_fields":["summary","key_insights","analysis_type"],"summary":{"text":"The report finds that while YouTube's algorithm primarily recommends Spanish-language news to both mainstream and fringe users, it serves conspiracy content to those who seek it at a lower rate than it serves lifestyle vlogger content to mainstream users. However, the fringe account received significantly more advertisements for Facebook and education, and a small number of recommendations related to weapons and military content.","ai_generated":true,"unit":"isdglo-000138#summary","date_modified":"2026-10-23"},"key_insights":[{"text":"The majority of recommended videos for both the mainstream and fringe accounts were aligned with their primary interest in Spanish-language news content.","ai_generated":true,"unit":"isdglo-000138#insight-1","date_modified":"2026-10-23"},{"text":"The algorithm recommended conspiracy videos to the fringe profile that had viewed such content, but at a lower rate than the mainstream profile was recommended its secondary interest of vlogger content. Specifically, 10% of videos recommended more than once to the fringe account were conspiracy-related, compared to over 30% of vlogger-related recommendations for the mainstream account.","ai_generated":true,"unit":"isdglo-000138#insight-2","date_modified":"2026-10-23"},{"text":"The fringe account received a small number of recommendations for weapons and military content, including one video on handguns and one on military airplanes, which the authors describe as a concerning trend.","ai_generated":true,"unit":"isdglo-000138#insight-3","date_modified":"2026-10-23"},{"text":"Advertisements differed substantially between the two profiles; the fringe account was served twice as many education ads (specifically data science and coding) and significantly more ads for Facebook than the mainstream account.","ai_generated":true,"unit":"isdglo-000138#insight-4","date_modified":"2026-10-23"},{"text":"The mainstream account was recommended significantly more videos related to religion, specifically evangelical Christianity, though the reason for this is unclear.","ai_generated":true,"unit":"isdglo-000138#insight-5","date_modified":"2026-10-23"},{"text":"The mainstream account received a higher total volume of recommendations, with 595 total videos and 12 ads, while the fringe account received 482 total videos and 24 ads.","ai_generated":true,"unit":"isdglo-000138#insight-6","date_modified":"2026-10-23"}],"climate_tags":[],"geography_discussed":[],"frame_perspective":[],"analysis_type":["algorithmic audit","comparative analysis"]},{"id":"isdglo-000063","url":"https://climateinsightsdirectory.org/document/isdglo-000063","title":"AMPLIFYING ANTISEMITISM: How Recommender Algorithms serve harmful content to children","publisher":{"name":"Institute for Strategic Dialogue","organization":"institute-for-strategic-dialogue"},"document_type":"report","language":"en","date_published":null,"source_url":"https://www.isdglobal.org/wp-content/uploads/2026/05/Amplifying-Antisemitism_How-Recommender-Algorithms-Serve-Harmful-Content-to-Children.pdf","generator":"Climate Insights Directory","generated_fields":["summary","key_insights","geography_discussed","analysis_type"],"summary":{"text":"The research demonstrates that TikTok's algorithms can nudge users from benign lifestyle content toward antisemitic conspiracy theories and far-right tropes within hours. On Rumble, the 'Editor's Picks' section provides more direct exposure to overt antisemitism, including slurs and calls to violence, exacerbated by a lack of age-verification requirements. The report highlights the role of interactive features like stickers and sounds in amplifying hate and calls for systemic regulatory interventions by Ofcom to address passive exposure pathways.","ai_generated":true,"unit":"isdglo-000063#summary","date_modified":"2026-10-23"},"key_insights":[{"text":"TikTok's recommender systems can rapidly lead minors to antisemitic content even without active searching; for example, a 15-year-old interested in male lifestyle influencers was served antisemitic conspiracy theories after only one hour on the platform.","ai_generated":true,"unit":"isdglo-000063#insight-1","date_modified":"2026-10-23"},{"text":"TikTok's algorithm creates pathways that escalate content from benign to harmful, where 'Culture & Lifestyle' content serves as a central entry point leading toward highly politicised and conspiratorial clusters.","ai_generated":true,"unit":"isdglo-000063#insight-2","date_modified":"2026-10-23"},{"text":"Interactive features on TikTok, specifically stickers, sounds, and comment sections, act as bridges that connect potentially benign videos to overt antisemitic content, including neo-Nazi imagery and Holocaust memes.","ai_generated":true,"unit":"isdglo-000063#insight-3","date_modified":"2026-10-23"},{"text":"There is a significant overlap between the 'manosphere' and antisemitism, where accounts prompted toward male lifestyle influencers were served misogynistic and antisemitic conspiracies, including the glorification of terrorist Elliot Rodger.","ai_generated":true,"unit":"isdglo-000063#insight-4","date_modified":"2026-10-23"},{"text":"Rumble's 'Editor's Picks' section exposes users to direct antisemitic slurs, conspiracy myths, and calls to violence immediately, unlike the gradual escalation seen on TikTok.","ai_generated":true,"unit":"isdglo-000063#insight-5","date_modified":"2026-10-23"},{"text":"Both TikTok and Rumble amplify persistent conspiracy narratives, such as those involving the Rothschilds, Jewish control of governments, and false claims that Jewish people were responsible for the 9/11 and October 7 attacks.","ai_generated":true,"unit":"isdglo-000063#insight-6","date_modified":"2026-10-23"},{"text":"Holocaust distortion and denial content, such as videos using coded phrases like '6 gorillion' or '271k' to contest the six million death toll, shows high levels of engagement with tens of thousands of likes and shares.","ai_generated":true,"unit":"isdglo-000063#insight-7","date_modified":"2026-10-23"},{"text":"Platform moderation is described as patchy and reactive; while some videos are removed, harmful stickers and coded sounds often remain, and Rumble's livestream features allow paid antisemitic GIFs to appear prominently.","ai_generated":true,"unit":"isdglo-000063#insight-8","date_modified":"2026-10-23"},{"text":"Rumble's non-account model represents a significant safeguarding gap because it allows minors to be recommended the same content as adults without requiring registration or age control.","ai_generated":true,"unit":"isdglo-000063#insight-9","date_modified":"2026-10-23"}],"climate_tags":[],"geography_discussed":["Canada","United States"],"frame_perspective":[],"analysis_type":["algorithmic audit","thematic analysis","policy analysis"]}]}