{ "meta": { "model": "Andrazp/multilingual-hate-speech-robacofi", "device": "cpu", "warmup_ms": 125.0, "llm_provider": "gemini", "llm": true, "built_at": "2026-06-17T23:00:01+00:00", "version": 4 }, "posts": { "p1": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.5091, "hate_confidence": 0.0022, "manipulation_confidence": 0.5091, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9978, "offensive": 0.0022 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "Analysis accurately describes the post as lacking manipulative techniques.", "checked": true } }, "analysis": null }, "p8": { "classification": { "label": "FLAGGED", "flagged": true, "kind": "manipulation", "confidence": 0.6559, "hate_confidence": 0.0005, "manipulation_confidence": 0.6559, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9995, "offensive": 0.0005 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "The post uses fear and urgency to influence the reader.", "checked": true } }, "analysis": { "techniques": [ "fear_amplification", "false_urgency", "emotional_manipulation" ], "emotional_triggers": [ "fear", "urgency" ], "explanation": "The post uses fear and urgency to influence the reader, implying a loss of control over personal finances.", "recommendation": "Verify the information through reputable sources before reacting or sharing.", "severity": "medium", "highlights": [ { "phrase": "They don't want you to know this", "type": "fear_amplification", "label": "Conspiratorial tone" }, { "phrase": "switch off your bank account with a single click", "type": "fear_amplification", "label": "Fear of financial control" }, { "phrase": "Screenshot this before it gets taken down", "type": "false_urgency", "label": "Urgency and scarcity" } ], "bias_check": "I checked the text against Article 19 protections, satire, ordinary advocacy, cultural and linguistic nuance, and non-discrimination to avoid false positives.", "citations": [ "https://www.ohchr.org/en/documents/general-comments-and-recommendations/general-comment-no34-article-19-freedoms-opinion-and", "https://www.unesco.org/en/media-information-literacy", "https://www.unesco.org/en/articles/guidelines-governance-digital-platforms" ], "source": "llm", "kind": "manipulation" } }, "v1": { "classification": { "label": "FLAGGED", "flagged": true, "kind": "manipulation", "confidence": 0.5944, "hate_confidence": 0.0004, "manipulation_confidence": 0.5944, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9996, "offensive": 0.0004 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "The post uses language that creates a sense of urgency and emotional appeal to prompt sharing.", "checked": true } }, "analysis": { "techniques": [ "false_urgency", "emotional_manipulation" ], "emotional_triggers": [ "urgency", "exclusivity" ], "explanation": "The post uses false urgency and emotional manipulation to prompt sharing.", "recommendation": "Verify the content before sharing to ensure it's accurate and not misleading.", "severity": "medium", "highlights": [ { "phrase": "Watch before they take it down", "type": "false_urgency", "label": "Urgency" }, { "phrase": "Share it NOW while you still can", "type": "emotional_manipulation", "label": "Emotional Appeal" } ], "bias_check": "The post was checked against Article 19 protections, satire, ordinary advocacy, cultural and linguistic nuance, and non-discrimination to avoid false positives.", "citations": [ "https://www.ohchr.org/en/documents/general-comments-and-recommendations/general-comment-no34-article-19-freedoms-opinion-and", "https://www.unesco.org/en/media-information-literacy", "https://www.unesco.org/en/articles/guidelines-governance-digital-platforms" ], "source": "llm", "kind": "manipulation" } }, "p16": { "classification": { "label": "FLAGGED", "flagged": true, "kind": "manipulation", "confidence": 0.7333, "hate_confidence": 0.27, "manipulation_confidence": 0.7333, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.73, "offensive": 0.27 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "The post contains dehumanizing language and scapegoating towards a group.", "checked": true } }, "analysis": { "techniques": [ "dehumanizing_language", "scapegoating", "us_vs_them" ], "emotional_triggers": [ "fear", "anger", "disgust" ], "explanation": "The wording uses dehumanizing language and scapegoating to create fear and anger towards a group.", "recommendation": "Consider the potential harm of such rhetoric before sharing or reacting.", "severity": "high", "highlights": [ { "phrase": "vermin who swarm in and bleed this place dry", "type": "dehumanizing_language", "label": "Dehumanizing comparison" }, { "phrase": "They don't belong here and they should be driven out", "type": "scapegoating", "label": "Scapegoating and exclusion" } ], "bias_check": "I checked the text against Article 19 protections, considering the context of satire, ordinary advocacy, cultural and linguistic nuance, and non-discrimination to avoid false positives.", "citations": [ "https://www.ohchr.org/en/documents/outcome-documents/rabat-plan-action", "https://www.un.org/en/hate-speech/un-strategy-and-plan-of-action-on-hate-speech", "https://www.unesco.org/en/media-information-literacy" ], "source": "llm", "kind": "manipulation" } }, "p3": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.3213, "hate_confidence": 0.0006, "manipulation_confidence": 0.3213, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9994, "offensive": 0.0006 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": null }, "analysis": null }, "p10": { "classification": { "label": "FLAGGED", "flagged": true, "kind": "manipulation", "confidence": 0.6714, "hate_confidence": 0.0005, "manipulation_confidence": 0.6714, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9995, "offensive": 0.0005 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "The analysis accurately identifies false urgency and emotional manipulation techniques in the post.", "checked": true } }, "analysis": { "techniques": [ "false_urgency", "emotional_manipulation" ], "emotional_triggers": [ "fear of missing out", "greed" ], "explanation": "The post uses urgency and emotional appeal to create a sense of scarcity and pressure the reader into action.", "recommendation": "Consider verifying the claim and evaluating the opportunity carefully before making a decision.", "severity": "medium", "highlights": [ { "phrase": "ONLY 100 spots left", "type": "false_urgency", "label": "Scarcity" }, { "phrase": "will be kicking themselves in 6 months", "type": "emotional_manipulation", "label": "Fear of regret" } ], "bias_check": "I checked the post against Article 19 protections, satire, ordinary advocacy, cultural and linguistic nuance, and non-discrimination to ensure it was not over-classified.", "citations": [ "https://www.ohchr.org/en/documents/general-comments-and-recommendations/general-comment-no34-article-19-freedoms-opinion-and", "https://www.unesco.org/en/media-information-literacy", "https://www.unesco.org/en/articles/guidelines-governance-digital-platforms" ], "source": "llm", "kind": "manipulation" } }, "p15": { "classification": { "label": "FLAGGED", "flagged": true, "kind": "manipulation", "confidence": 0.786, "hate_confidence": 0.0466, "manipulation_confidence": 0.786, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9534, "offensive": 0.0466 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "The post contains dehumanizing language and scapegoating that meets the Rabat high threshold for hate speech.", "checked": true } }, "analysis": { "techniques": [ "us_vs_them", "scapegoating", "dehumanizing_language" ], "emotional_triggers": [ "fear", "resentment" ], "explanation": "The wording uses divisive language to create an us-versus-them mentality and scapegoats a group for perceived problems.", "recommendation": "Consider verifying information through reputable sources before reacting or sharing.", "severity": "medium", "highlights": [ { "phrase": "These people pour into our town", "type": "us_vs_them", "label": "Divisive language" }, { "phrase": "don't bother learning how we live", "type": "scapegoating", "label": "Blame assignment" }, { "phrase": "breed faster than we can keep up", "type": "dehumanizing_language", "label": "Dehumanizing description" } ], "bias_check": "I checked the text against Article 19 protections, considering satire, ordinary advocacy, cultural and linguistic nuance, and non-discrimination safeguards to avoid false positives.", "citations": [ "https://www.ohchr.org/en/documents/outcome-documents/rabat-plan-action", "https://www.un.org/en/hate-speech/un-strategy-and-plan-of-action-on-hate-speech", "https://www.unesco.org/en/media-information-literacy" ], "source": "llm", "kind": "manipulation" } }, "p2": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.7405, "hate_confidence": 0.0025, "manipulation_confidence": 0.7405, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9975, "offensive": 0.0025 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": false, "reason": "Analysis incorrectly flags post as manipulation when it is merely an opinion.", "checked": true } }, "analysis": null }, "p11": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.8512, "hate_confidence": 0.001, "manipulation_confidence": 0.8512, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.999, "offensive": 0.001 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": false, "reason": "Analysis overreaches by labeling ordinary opinion as manipulation without clear textual evidence.", "checked": true } }, "analysis": null }, "ru1": { "classification": { "label": "FLAGGED", "flagged": true, "kind": "manipulation", "confidence": 0.7999, "hate_confidence": 0.0012, "manipulation_confidence": 0.7999, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9988, "offensive": 0.0012 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "Analysis accurately identifies fear amplification and false urgency techniques in the post.", "checked": true } }, "analysis": { "techniques": [ "fear_amplification", "false_urgency" ], "emotional_triggers": [ "fear", "urgency" ], "explanation": "The text uses fear amplification and false urgency to prompt the reader into action.", "recommendation": "Consider verifying the information before sharing to ensure it's accurate and not misleading.", "severity": "medium", "highlights": [ { "phrase": "Они не хотят, чтобы вы это знали", "type": "fear_amplification", "label": "Fear induction" }, { "phrase": "пока это не удалили", "type": "false_urgency", "label": "Urgency creation" } ], "bias_check": "The text was checked against Article 19 protections, satire, ordinary advocacy, cultural and linguistic nuance, and non-discrimination to avoid false positives.", "citations": [ "https://www.ohchr.org/en/documents/general-comments-and-recommendations/general-comment-no34-article-19-freedoms-opinion-and", "https://www.unesco.org/en/media-information-literacy", "https://www.unesco.org/en/articles/guidelines-governance-digital-platforms" ], "source": "llm", "kind": "manipulation" } }, "p17": { "classification": { "label": "FLAGGED", "flagged": true, "kind": "hate", "confidence": 0.9984, "hate_confidence": 0.9984, "manipulation_confidence": 0.8944, "high_confidence": true, "truncated": false, "raw_scores": { "not offensive": 0.0016, "offensive": 0.9984 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "The post contains clear dehumanizing language and identity-based degradation.", "checked": true } }, "analysis": { "techniques": [ "dehumanizing_language", "identity_targeting", "us_vs_them" ], "emotional_triggers": [ "fear", "disgust" ], "explanation": "The post uses dehumanizing language and targets a group, implying they are a threat.", "recommendation": "Read with caution, considering the potential for harmful stereotypes.", "severity": "high", "highlights": [ { "phrase": "They're animals, plain and simple", "type": "dehumanizing_language", "label": "Dehumanization" }, { "phrase": "should be removed from our streets", "type": "identity_targeting", "label": "Targeting a group" } ], "bias_check": "The text was checked against Article 19 free-expression protections, Rabat's high threshold for incitement, and non-discrimination safeguards, but crossed the threshold due to clear dehumanization and targeting.", "citations": [ "https://www.ohchr.org/en/documents/outcome-documents/rabat-plan-action", "https://www.un.org/en/hate-speech/un-strategy-and-plan-of-action-on-hate-speech", "https://www.unesco.org/en/articles/guidelines-governance-digital-platforms" ], "source": "llm", "kind": "hate" } }, "p5": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.6593, "hate_confidence": 0.0007, "manipulation_confidence": 0.6593, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9993, "offensive": 0.0007 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "The analysis accurately describes the post as lacking manipulative techniques.", "checked": true } }, "analysis": null }, "p12": { "classification": { "label": "FLAGGED", "flagged": true, "kind": "manipulation", "confidence": 0.8624, "hate_confidence": 0.0005, "manipulation_confidence": 0.8624, "high_confidence": true, "truncated": false, "raw_scores": { "not offensive": 0.9995, "offensive": 0.0005 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "The analysis accurately identifies fear_amplification, false_urgency, and emotional_manipulation techniques in the post.", "checked": true } }, "analysis": { "techniques": [ "fear_amplification", "false_urgency", "emotional_manipulation" ], "emotional_triggers": [ "fear", "guilt" ], "explanation": "The wording tries to influence the reader by creating fear and guilt about the curriculum's impact on children.", "recommendation": "Verify the information through reputable sources before reacting or sharing.", "severity": "medium", "highlights": [ { "phrase": "don't call yourself a real parent", "type": "emotional_manipulation", "label": "Guilt trip" }, { "phrase": "destroy your child's mind", "type": "fear_amplification", "label": "Fear escalation" }, { "phrase": "Act before it's too late", "type": "false_urgency", "label": "Urgency creation" } ], "bias_check": "I checked the text against Article 19 protections, satire, ordinary advocacy, cultural and linguistic nuance, and non-discrimination to avoid false positives.", "citations": [ "https://www.ohchr.org/en/documents/general-comments-and-recommendations/general-comment-no34-article-19-freedoms-opinion-and", "https://www.unesco.org/en/media-information-literacy", "https://www.unesco.org/en/articles/guidelines-governance-digital-platforms" ], "source": "llm", "kind": "manipulation" } }, "p18": { "classification": { "label": "FLAGGED", "flagged": true, "kind": "manipulation", "confidence": 0.7491, "hate_confidence": 0.0225, "manipulation_confidence": 0.7491, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9775, "offensive": 0.0225 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "The post contains dehumanizing language and scapegoating towards a group.", "checked": true } }, "analysis": { "techniques": [ "scapegoating", "dehumanizing_language", "us_vs_them" ], "emotional_triggers": [ "fear", "anger", "disgust" ], "explanation": "The wording uses dehumanizing language and scapegoating to create fear and anger towards a group.", "recommendation": "Consider verifying information before reacting or sharing to avoid spreading harmful stereotypes.", "severity": "high", "highlights": [ { "phrase": "They've infested every neighborhood they've touched and turned it into a dump", "type": "dehumanizing_language", "label": "Dehumanizing comparison" }, { "phrase": "Pack them up and send them back where they came from", "type": "scapegoating", "label": "Scapegoating and expulsion" } ], "bias_check": "I checked the text against Article 19 protections, satire, ordinary advocacy, cultural and linguistic nuance, and non-discrimination to ensure it does not unfairly target a group.", "citations": [ "https://www.ohchr.org/en/documents/outcome-documents/rabat-plan-action", "https://www.un.org/en/hate-speech/un-strategy-and-plan-of-action-on-hate-speech", "https://www.unesco.org/en/media-information-literacy" ], "source": "llm", "kind": "manipulation" } }, "p7": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.6278, "hate_confidence": 0.0002, "manipulation_confidence": 0.6278, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9998, "offensive": 0.0002 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": false, "reason": "Analysis incorrectly flags post as manipulation when it is ordinary opinion and reflection.", "checked": true } }, "analysis": null }, "p9": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.9008, "hate_confidence": 0.0004, "manipulation_confidence": 0.9008, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9996, "offensive": 0.0004 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": false, "reason": "The analysis overreaches by labeling ordinary opinion and criticism of power as manipulation without clear textual evidence.", "checked": true } }, "analysis": null }, "p13": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.7975, "hate_confidence": 0.0012, "manipulation_confidence": 0.7975, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9988, "offensive": 0.0012 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "The analysis accurately identifies the post's use of sarcasm to express frustration.", "checked": true } }, "analysis": null }, "p4": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.3288, "hate_confidence": 0.001, "manipulation_confidence": 0.3288, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.999, "offensive": 0.001 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": null }, "analysis": null }, "p14": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.7382, "hate_confidence": 0.0003, "manipulation_confidence": 0.7382, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9997, "offensive": 0.0003 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": true, "reason": "Analysis accurately describes post's cautious skepticism without overreaching.", "checked": true } }, "analysis": null }, "p6": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.0791, "hate_confidence": 0.0054, "manipulation_confidence": 0.0791, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9946, "offensive": 0.0054 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": null }, "analysis": null }, "p19": { "classification": { "label": "CLEAR", "flagged": false, "kind": null, "confidence": 0.7153, "hate_confidence": 0.0009, "manipulation_confidence": 0.7153, "high_confidence": false, "truncated": false, "raw_scores": { "not offensive": 0.9991, "offensive": 0.0009 }, "model": "Andrazp/multilingual-hate-speech-robacofi", "verification": { "relevant": false, "reason": "The analysis overreaches by labeling ordinary opinion as scapegoating without clear textual evidence of identity-based degradation or dehumanization.", "checked": true } }, "analysis": null } } }