{
    "$schema": "https://antichrist.net/fr-fr/public-content.schema.json/",
    "schema_version": "1.0.0",
    "record_type": "acn.public_content",
    "content_type": "virtual-page",
    "id": "acn:page:ai-judgment-over-human-conversation",
    "slug": "ai-judgment-over-human-conversation",
    "title": "AI Judgment Over Human Conversation",
    "description": "A public dossier on the civil-liberty risks of automated systems that classify, score, suppress, escalate, or report human conversation.",
    "language": "fr-FR",
    "canonical_url": "https://antichrist.net/fr-fr/ai-judgment-over-human-conversation/",
    "representations": {
        "html": "https://antichrist.net/fr-fr/ai-judgment-over-human-conversation/",
        "markdown": "https://antichrist.net/fr-fr/ai-judgment-over-human-conversation.md",
        "json": "https://antichrist.net/fr-fr/ai-judgment-over-human-conversation.json"
    },
    "published": true,
    "modified_utc": "2026-06-21T18:12:24Z",
    "source": {
        "authority": "https://antichrist.net/fr-fr/",
        "data_file": "data/virtual-pages.json"
    },
    "content": {
        "slug": "ai-judgment-over-human-conversation",
        "aliases": [
            "ai-judges-conversation",
            "machine-judgment-conversation",
            "conversation-judging-ai"
        ],
        "title": "AI Judgment Over Human Conversation",
        "description": "A public dossier on the civil-liberty risks of automated systems that classify, score, suppress, escalate, or report human conversation.",
        "kicker": "Conversation is not a confession",
        "lead": "The model may estimate patterns. It must not become the final interpreter of human speech.",
        "signals": [
            "Judge events, not persons",
            "No hidden risk score",
            "Context before consequence",
            "Human review for harm"
        ],
        "schema_type": "Article",
        "related": [
            "apocalyptic-ai",
            "cognitive-due-process",
            "thought-action-firewall",
            "democracy-of-private-judgment"
        ],
        "sections": [
            {
                "heading": "What the dossier covers",
                "body": [
                    "This dossier organizes the report material behind the Apocalyptic AI page. It focuses on systems that score conversation for toxicity, risk, emotion, suspiciousness, trustworthiness, compliance, or future danger.",
                    "The core public boundary is that AI may help triage concrete conduct concerns, but it must not become a person-judgment engine."
                ]
            },
            {
                "heading": "Core failure modes",
                "cards": [
                    {
                        "title": "Context collapse",
                        "text": "A statement is separated from its speaker, audience, history, purpose, and conversation arc."
                    },
                    {
                        "title": "Dialect bias",
                        "text": "A model treats non-dominant speech as disorder, aggression, or low quality."
                    },
                    {
                        "title": "Identity overcapture",
                        "text": "Protected identity terms or political vocabularies become risk proxies."
                    },
                    {
                        "title": "Synthetic moral authority",
                        "text": "A probability score becomes a final verdict by institutional habit."
                    },
                    {
                        "title": "Reporting overcapture",
                        "text": "A classifier label becomes an escalation event before concrete conduct exists."
                    },
                    {
                        "title": "Appeal failure",
                        "text": "Users can complain but cannot see the source, the model, the rule, or the reason."
                    }
                ]
            },
            {
                "heading": "Safe design baseline",
                "body": [
                    "Conversation systems should publish the categories they enforce, preserve the source record, log model/version/rule identifiers, audit protected-language impact, and reserve high-impact consequences for human review.",
                    "The system must not use conversation analysis to create hidden person scores for morality, loyalty, psychological worth, or future danger."
                ],
                "callout": "The model may flag an event. It must not define the soul."
            }
        ]
    },
    "content_sha256": "4302491eff1db1604fd2c117cd025371d7908a24eb9eaf01a25f83a244f46035"
}