{"passport":{"unfragile":{"@version":"1.0","version":"2026-05","artifact":{"id":"tool_ai-detector","slug":"ai-detector","name":"AI Detector","type":"product","url":"https://writehuman.ai","page_url":"https://unfragile.ai/ai-detector","categories":["text-writing"],"tags":[],"pricing":{"model":"paid","free":false,"starting_price":null},"status":"active","verified":false},"capabilities":[{"id":"tool_ai-detector__cap_0","uri":"capability://safety.moderation.single.text.authenticity.classification","name":"single-text-authenticity-classification","description":"Analyzes submitted text through a trained neural classifier to determine probability of AI generation, returning a confidence score and binary classification (AI-generated vs human-written). The system processes input text through feature extraction layers that identify statistical patterns, linguistic markers, and stylistic anomalies characteristic of LLM outputs, then applies a decision threshold to produce instant results without requiring API calls or external model inference.","intents":["I need to quickly check if a student essay was written by ChatGPT or a human","I want to verify whether a submitted article is original human writing before publishing","I need a fast preliminary screening to flag suspicious content for manual review"],"best_for":["educators grading assignments in bulk who need sub-second feedback","content moderators performing initial triage on user-submitted text","small teams without budget for enterprise detection solutions"],"limitations":["Detection accuracy drops significantly against GPT-4o, Claude 3.5, and other frontier models—false negative rate increases above 30% on sophisticated outputs","Inconsistent performance across writing domains (technical writing shows higher false positives than narrative prose)","No adaptive learning—model weights remain static and cannot be fine-tuned to specific writing styles or domains","Vulnerable to simple obfuscation techniques like synonym replacement or sentence restructuring"],"requires":["Web browser with JavaScript enabled","Plain text input (no file upload preprocessing for PDFs or Word documents)","Active internet connection for backend inference"],"input_types":["plain text (copy-paste)","text strings up to ~5000 words per submission"],"output_types":["confidence score (0-100 percentage)","binary classification label (AI-generated / Human-written)","optional explanation text"],"categories":["safety-moderation","text-analysis"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"tool_ai-detector__cap_1","uri":"capability://data.processing.analysis.batch.text.processing.with.csv.export","name":"batch-text-processing-with-csv-export","description":"Accepts multiple text submissions (either pasted individually or uploaded as structured data) and processes them sequentially through the authenticity classifier, aggregating results into a downloadable CSV or JSON report with per-document scores, classifications, and metadata. The system queues submissions and distributes inference across available compute resources, though without true parallel processing—each document is classified serially with results cached to prevent duplicate analysis.","intents":["I need to screen 50 student essays at once and export results to a spreadsheet for grading","I want to analyze a dataset of user-generated content and identify which pieces are likely AI-written","I need to generate a compliance report showing which submissions passed authenticity checks"],"best_for":["educators managing large class sections (50+ students)","content platforms performing moderation at scale (100s of submissions per day)","researchers studying AI detection accuracy across document collections"],"limitations":["No true parallelization—batch processing is sequential, making 100+ document batches slow (estimated 1-2 minutes for 50 documents)","CSV export lacks granular metadata (no per-sentence confidence scores, only document-level aggregates)","No scheduled batch processing—all submissions must be uploaded in single session","File size limits prevent processing of very large documents (>10,000 words per submission may timeout)"],"requires":["Web browser with file upload capability","CSV or plain text input format","Active WriteHuman account (free tier may have daily submission limits)"],"input_types":["CSV with text column","plain text files (one submission per line or per file)","pasted text blocks separated by delimiters"],"output_types":["CSV export with columns: [document_id, text_preview, ai_score, classification, timestamp]","JSON report with nested metadata","downloadable report file"],"categories":["data-processing-analysis","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"tool_ai-detector__cap_2","uri":"capability://safety.moderation.confidence.score.interpretation.with.thresholds","name":"confidence-score-interpretation-with-thresholds","description":"Returns a numerical confidence score (typically 0-100 scale) representing the model's certainty that text is AI-generated, paired with interpretive guidance on what different score ranges mean. The system applies configurable decision thresholds (e.g., >75 = likely AI, 25-75 = ambiguous, <25 = likely human) and may provide explanatory text highlighting specific linguistic features that contributed to the classification, though the exact feature attribution mechanism is not transparent.","intents":["I want to know not just if text is AI, but HOW CONFIDENT the detector is in that judgment","I need to set a custom threshold for my use case (e.g., flag anything >60% as suspicious)","I want to understand WHY the detector flagged a submission so I can make an informed decision"],"best_for":["educators who want to use detection as a starting point for conversation, not final verdict","content teams needing to balance false positives against false negatives","researchers studying detection reliability across different confidence bands"],"limitations":["Threshold customization is not exposed in UI—users cannot adjust decision boundaries for their specific domain","Feature attribution (which linguistic markers triggered the classification) is opaque or absent—no explainability layer","Confidence scores are not calibrated—a 75% score does not reliably correspond to 75% actual probability of AI generation","No confidence intervals or uncertainty quantification—single point estimate without bounds"],"requires":["Understanding of what confidence scores mean in context of false positive/negative rates","Domain knowledge to interpret ambiguous scores (25-75 range)"],"input_types":["text submission (same as single-text-authenticity-classification)"],"output_types":["numerical confidence score (0-100)","categorical interpretation (e.g., 'Likely AI', 'Ambiguous', 'Likely Human')","optional explanatory text or feature highlights"],"categories":["safety-moderation","data-processing-analysis"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"tool_ai-detector__cap_3","uri":"capability://safety.moderation.multi.language.detection.support","name":"multi-language-detection-support","description":"Extends the authenticity classifier to handle text in multiple languages beyond English, applying language-specific feature extraction and classification models. The system detects input language automatically (or accepts explicit language specification) and routes text to the appropriate language-trained classifier, though support is limited to a subset of high-resource languages and performance degrades for low-resource or code-mixed inputs.","intents":["I need to check if a Spanish-language essay was written by a student or generated by ChatGPT","I want to screen user submissions in multiple languages without running separate tools","I need to verify authenticity of content in my non-English market"],"best_for":["international educators working with multilingual classrooms","global content platforms with submissions in 5+ languages","non-English-speaking regions where AI detection is critical but tools are scarce"],"limitations":["Language support is limited—likely covers only 5-10 major languages (English, Spanish, French, German, Chinese, Japanese) with no documentation of full list","Detection accuracy varies significantly by language—English models are most mature, others show 10-20% lower F1 scores","No support for code-mixed text (e.g., Spanglish, Hinglish)—language detection fails and classification becomes unreliable","Automatic language detection can misidentify short texts or texts with mixed scripts"],"requires":["Text in supported language (no explicit language list provided)","Optional: explicit language specification to bypass auto-detection"],"input_types":["plain text in supported languages","language code specification (e.g., 'es' for Spanish, 'fr' for French)"],"output_types":["confidence score (0-100)","classification label in detected/specified language","detected language code"],"categories":["safety-moderation","text-generation-language"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"tool_ai-detector__cap_4","uri":"capability://safety.moderation.plagiarism.detection.integration.optional","name":"plagiarism-detection-integration-optional","description":"May integrate with or reference plagiarism detection capabilities (either native or via third-party APIs like Turnitin) to provide a combined authenticity check—flagging both AI-generated content AND plagiarized human content in a single analysis. The integration approach is unclear from available documentation, but likely involves either sequential API calls or a unified scoring interface that combines AI detection confidence with plagiarism match percentages.","intents":["I want to check if a submission is AI-generated OR plagiarized in one step","I need a comprehensive authenticity report that covers both AI and plagiarism risks","I want to avoid false positives from AI detection by cross-checking against plagiarism data"],"best_for":["educational institutions with existing plagiarism detection workflows","content platforms needing comprehensive authenticity screening","compliance teams managing content provenance"],"limitations":["Integration status is unclear—may not be available in all tiers or may require separate subscription","No documentation of how AI detection and plagiarism scores are combined or weighted","Plagiarism detection accuracy depends on third-party provider (if integrated), not WriteHuman's own implementation","May add significant latency if plagiarism check requires external API call (2-5 seconds additional)"],"requires":["Plagiarism detection subscription (if not bundled)","API credentials for third-party plagiarism service (if applicable)"],"input_types":["plain text (same as core detection)"],"output_types":["combined authenticity report with AI score + plagiarism percentage","unified risk classification (e.g., 'High Risk: AI + Plagiarism', 'Medium Risk: AI Only')"],"categories":["safety-moderation","tool-use-integration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"tool_ai-detector__cap_5","uri":"capability://tool.use.integration.api.endpoint.for.programmatic.access","name":"api-endpoint-for-programmatic-access","description":"Exposes the authenticity classifier as a REST API endpoint, allowing developers to integrate AI detection into custom applications, LMS platforms, or content management systems without using the web UI. The API likely accepts JSON payloads with text content and returns structured JSON responses with confidence scores and classifications, though rate limiting, authentication mechanisms, and SLA guarantees are not documented.","intents":["I want to embed AI detection into my LMS so students get instant feedback on submissions","I need to call the detector from my backend service for every user-generated post","I want to build a custom workflow that chains AI detection with other content moderation tools"],"best_for":["developers building custom integrations with educational platforms","content platforms needing server-side detection without UI overhead","teams automating content moderation pipelines"],"limitations":["API documentation is minimal or absent—no public OpenAPI spec, rate limits, or SLA documentation","Authentication mechanism unclear—likely API key-based, but key management and rotation procedures are undocumented","Rate limiting is likely enforced but thresholds are unknown—could be 10 req/min or 1000 req/min","No batch endpoint—API likely requires one request per text submission, making large-scale processing inefficient","No webhook support for asynchronous processing—all requests are synchronous"],"requires":["API key (provisioned via WriteHuman account)","HTTP client library (curl, requests, axios, etc.)","Developer account with API access enabled"],"input_types":["JSON payload with 'text' field","optional: language code, custom threshold parameters"],"output_types":["JSON response with 'score', 'classification', 'confidence' fields","HTTP status codes (200 for success, 429 for rate limit, 401 for auth failure)"],"categories":["tool-use-integration","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"tool_ai-detector__cap_6","uri":"capability://safety.moderation.writing.style.fingerprinting.for.consistency.checks","name":"writing-style-fingerprinting-for-consistency-checks","description":"Analyzes stylistic patterns within submitted text (vocabulary diversity, sentence structure, punctuation habits, tone consistency) to detect sudden shifts that might indicate AI generation or content splicing. The system builds a statistical profile of the author's baseline writing style from the submitted text itself or from a reference corpus, then flags sections that deviate significantly from that profile as potentially AI-generated or plagiarized.","intents":["I want to detect if a student's essay suddenly shifts to a different writing style midway through","I need to identify sections of a document that don't match the author's typical voice","I want to catch cases where a student pasted AI-generated paragraphs into their own writing"],"best_for":["educators familiar with individual student writing styles","content teams managing author consistency across publications","plagiarism investigators looking for spliced content"],"limitations":["Requires sufficient text length to establish baseline style (likely 500+ words)—short submissions cannot be analyzed","Vulnerable to false positives when authors intentionally vary style (e.g., formal introduction vs casual conclusion)","Cannot distinguish between 'AI-generated' and 'written by a different human'—style shifts alone are ambiguous","No per-section confidence scores—flagged sections lack quantified risk levels","Ineffective against AI models trained on specific author styles (e.g., GPT fine-tuned on student's previous essays)"],"requires":["Minimum text length (500+ words recommended)","Optional: reference corpus of author's previous writing for baseline comparison"],"input_types":["plain text document","optional: reference texts for style baseline"],"output_types":["style consistency score (0-100)","flagged sections with deviation percentages","style profile summary (vocabulary richness, sentence length distribution, etc.)"],"categories":["safety-moderation","data-processing-analysis"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"tool_ai-detector__cap_7","uri":"capability://automation.workflow.user.account.management.with.submission.history","name":"user-account-management-with-submission-history","description":"Provides user authentication and account management, allowing users to create accounts, log in, and maintain a history of previous text submissions and their detection results. The system stores submission metadata (timestamp, text preview, scores, classifications) in a user-accessible dashboard, enabling users to track detection patterns over time and compare results across multiple submissions without re-running analysis.","intents":["I want to keep a record of all the essays I've checked so I can reference them later","I need to see trends in my class's AI usage over the semester","I want to compare detection results across multiple student submissions to identify patterns"],"best_for":["educators managing ongoing classes with repeated submissions","content teams tracking moderation history for compliance","researchers studying detection accuracy across time"],"limitations":["Submission history retention period is unclear—may be limited to 30-90 days or require paid tier for longer retention","No bulk export of historical data—users must download results individually or via batch export","Privacy implications unclear—stored submissions may be retained for model improvement without explicit consent","No multi-user account sharing—each educator must maintain separate account, limiting team collaboration"],"requires":["Email address for account creation","Password (or SSO integration if available)","Active internet connection"],"input_types":["login credentials","account profile information"],"output_types":["user dashboard with submission history","filterable/sortable submission list","historical trend charts (optional)"],"categories":["automation-workflow","memory-knowledge"],"confidence":0.5,"matches":0,"success_rate":0}],"trust":{"score":41,"verified":false,"data_access_risk":"high","permissions":["Web browser with JavaScript enabled","Plain text input (no file upload preprocessing for PDFs or Word documents)","Active internet connection for backend inference","Web browser with file upload capability","CSV or plain text input format","Active WriteHuman account (free tier may have daily submission limits)","Understanding of what confidence scores mean in context of false positive/negative rates","Domain knowledge to interpret ambiguous scores (25-75 range)","Text in supported language (no explicit language list provided)","Optional: explicit language specification to bypass auto-detection"],"failure_modes":["Detection accuracy drops significantly against GPT-4o, Claude 3.5, and other frontier models—false negative rate increases above 30% on sophisticated outputs","Inconsistent performance across writing domains (technical writing shows higher false positives than narrative prose)","No adaptive learning—model weights remain static and cannot be fine-tuned to specific writing styles or domains","Vulnerable to simple obfuscation techniques like synonym replacement or sentence restructuring","No true parallelization—batch processing is sequential, making 100+ document batches slow (estimated 1-2 minutes for 50 documents)","CSV export lacks granular metadata (no per-sentence confidence scores, only document-level aggregates)","No scheduled batch processing—all submissions must be uploaded in single session","File size limits prevent processing of very large documents (>10,000 words per submission may timeout)","Threshold customization is not exposed in UI—users cannot adjust decision boundaries for their specific domain","Feature attribution (which linguistic markers triggered the classification) is opaque or absent—no explainability layer","builder identity is not verified yet","no observed match outcomes yet"],"rank_breakdown":{"adoption":0.31666666666666665,"quality":0.67,"ecosystem":0.35,"match_graph":0.25,"freshness":0.75,"weights":{"adoption":0.25,"quality":0.25,"ecosystem":0.1,"match_graph":0.35,"freshness":0.05}},"observed_outcomes":{"matches":0,"success_rate":0,"avg_confidence":0,"top_intents":[],"last_matched_at":null},"maintenance":{"status":"active","updated_at":"2026-05-24T12:16:29.132Z","last_scraped_at":"2026-04-05T13:23:42.562Z","last_commit":null},"community":{"stars":null,"forks":null,"weekly_downloads":null,"model_downloads":null,"model_likes":null}},"distribution":{"claim_url":"https://unfragile.ai/submit?claim=ai-detector","compare_url":"https://unfragile.ai/compare?artifact=ai-detector"}},"signature":"HwImT9n6XtNgdcopQ+s/2DpjVDYtfE2K3sGpayTeirQZ7FwTxFaGNLBBvOQjRlfM5UJn35on6+LjrphF0E7kBQ==","signedAt":"2026-06-21T01:13:23.913Z","signedBy":"unfragile.ai","version":1},"_links":{"self":"https://unfragile.ai/api/v1/passport/ai-detector","artifact":"https://unfragile.ai/ai-detector","verify":"https://unfragile.ai/api/v1/verify?slug=ai-detector","publicKey":"https://unfragile.ai/api/v1/trust-passport-public-key","spec":"https://unfragile.ai/trust","schema":"https://unfragile.ai/schema.json","docs":"https://unfragile.ai/docs"}}