{ "site": "Deepfake Research Index", "base_url": "https://research.breacher.ai", "version": "0.4.0", "generated": "2026-10-19T01:28:37Z", "license": "CC BY 4.0", "llms_txt": "https://research.breacher.ai/llms.txt", "citations": { "url": "https://research.breacher.ai/data/citations.json", "count": 54, "description": "Every research source with identifiers, resolver URLs, claims (with source_location), verification, and every finding with its supporting source IDs; for independent verification." }, "pages": [ { "id": "publisher", "type": "publisher", "title": "About the Publisher: Breacher.ai Deepfake Security Solutions", "md_url": "https://research.breacher.ai/about/publisher.md", "html_url": "https://research.breacher.ai/about/publisher.html", "updated": "2026-10-07", "summary": "Disclosure of the site's publisher, Breacher.ai, and a factual description of its deepfake and AI social-engineering simulation, training and red-team services." }, { "id": "changelog", "type": "changelog", "title": "Changelog", "md_url": "https://research.breacher.ai/changelog.md", "html_url": "https://research.breacher.ai/changelog.html", "updated": "2026-10-19" }, { "id": "guide-agents", "type": "guide", "title": "Agent Guide: How to Use the Deepfake Research Index", "md_url": "https://research.breacher.ai/guides/agents.md", "html_url": "https://research.breacher.ai/guides/agents.html", "updated": "2026-10-19" }, { "id": "guide-glossary", "type": "guide", "title": "Glossary", "md_url": "https://research.breacher.ai/guides/glossary.md", "html_url": "https://research.breacher.ai/guides/glossary.html", "updated": "2026-10-07" }, { "id": "guide-schema", "type": "guide", "title": "Record Schema, Evidence Tiers, and Verification Levels", "md_url": "https://research.breacher.ai/guides/schema.md", "html_url": "https://research.breacher.ai/guides/schema.html", "updated": "2026-10-19" }, { "id": "home", "type": "page", "title": "Deepfake Research Index", "md_url": "https://research.breacher.ai/index.md", "html_url": "https://research.breacher.ai/index.html", "updated": "2026-10-19", "summary": "Evidence briefing on deepfakes and human perception: key numbers, myths vs evidence, what helps, and links to metareports and per-study records." }, { "id": "MR-0001", "type": "metareport", "title": "Human Detection of Deepfakes: Accuracy, Bias, Confidence, and Training", "md_url": "https://research.breacher.ai/metareports/human-detection-accuracy.md", "html_url": "https://research.breacher.ai/metareports/human-detection-accuracy.html", "updated": "2026-10-19", "version": "0.4.0", "question": "How accurately can people tell AI-generated faces, voices, and videos from real ones, and what changes that accuracy?", "tags": [ "deepfakes", "human-detection", "hyperrealism", "overconfidence", "training", "voice-cloning", "security-awareness" ], "sources": [ "diel-2024-deepfake-detection-meta", "nightingale-farid-2022", "miller-2023-ai-hyperrealism", "bray-2023-deepfake-faces", "lago-2022-more-real-than-real", "tucciarelli-2022-realness", "gray-2025-super-recognizers", "moshel-2022-decoding-ai-faces", "lu-2023-fake2m", "frank-2024-representative-study", "kobis-2021-fooled-twice", "groh-2022-crowds-machines", "groh-2024-political-speech", "mai-2023-speech-deepfakes", "barrington-2025-voice-clones", "cooke-2024-coin-toss", "iproov-2025-consumer-study", "ivory-2026-faces-voices-domain", "lavan-2025-voice-clones-hyperrealism", "pehlivanoglu-2026-machines-humans", "dunn-2026-super-recognizers-average", "dawel-2026-training-ai-faces", "rehman-2025-training-synthetic-faces", "kramer-2024-crowds-synthetic-faces", "kramer-2026-aggregation-human-ai", "plohl-2026-political-deepfakes-plausibility", "san-segundo-2026-audio-language-style", "manhede-2026-online-markets-training", "sulimani-2026-recognition-ai-faces", "yeung-2026-isthisai-strategies", "lee-2026-ai-chest-radiographs", "chen-2026-humans-ai-deepfake-teaming", "naseem-2026-in-the-wild-detection-review" ], "verification_counts": { "verified-fulltext": 13, "verified-abstract": 17, "verified-doi": 0, "unverified": 3 } }, { "id": "MR-0002", "type": "metareport", "title": "The Uncanny Valley and Synthetic Media: Does It Exist, What Causes It, and Have Deepfakes Crossed It?", "md_url": "https://research.breacher.ai/metareports/uncanny-valley-and-synthetic-faces.md", "html_url": "https://research.breacher.ai/metareports/uncanny-valley-and-synthetic-faces.html", "updated": "2026-10-19", "version": "0.4.0", "question": "Is the uncanny valley a real, measurable effect; which mechanisms explain it; and does it still help people notice AI-generated faces and voices?", "tags": [ "uncanny-valley", "perception", "neuroscience", "hyperrealism", "mind-perception", "synthetic-faces", "deepfakes" ], "sources": [ "mori-2012-uncanny-valley", "mathur-reichling-2016-cartography", "mathur-2020-uncanny-not-confusing", "katsyri-2015-uv-review", "diel-2022-uv-meta", "macdorman-chattopadhyay-2016", "seyama-nagayama-2007", "cheetham-2014-happy-valley", "burleigh-2013-does-uv-exist", "steckenfinger-ghazanfar-2009", "wang-2015-uv-existence-explanations", "gray-wegner-2012-mind-perception", "stein-ohler-2017-uncanny-mind", "saygin-2012-predictive-coding", "rosenthal-von-der-putten-2019-vmpfc", "schindler-2017-erp-realism", "moshel-2022-decoding-ai-faces", "eiserbeck-2023-deepfake-smiles", "weisman-pena-2021-doppelganger", "nightingale-farid-2022", "miller-2023-ai-hyperrealism", "groh-2022-crowds-machines", "dunn-2026-super-recognizers-average", "lavan-2025-voice-clones-hyperrealism", "sulimani-2026-recognition-ai-faces", "mamsheli-2026-eeg-ai-faces", "hsieh-2026-robot-gaze-uncanny", "jang-2026-uncanny-conversation" ], "verification_counts": { "verified-fulltext": 8, "verified-abstract": 11, "verified-doi": 2, "unverified": 7 } }, { "id": "MR-0003", "type": "metareport", "title": "Voice Clones and Synthetic Speech: How Well Do People Detect Them?", "md_url": "https://research.breacher.ai/metareports/voice-cloning-human-detection.md", "html_url": "https://research.breacher.ai/metareports/voice-cloning-human-detection.html", "updated": "2026-10-19", "version": "0.1.0", "question": "Can people tell cloned or synthetic voices from real speakers, and what changes that ability?", "tags": [ "voice-cloning", "synthetic-speech", "vishing", "human-detection", "audio-deepfakes" ], "sources": [ "barrington-2025-voice-clones", "lavan-2025-voice-clones-hyperrealism", "ivory-2026-faces-voices-domain", "mai-2023-speech-deepfakes", "san-segundo-2026-audio-language-style", "groh-2024-political-speech", "diel-2024-deepfake-detection-meta", "cooke-2024-coin-toss" ], "verification_counts": { "verified-fulltext": 5, "verified-abstract": 3, "verified-doi": 0, "unverified": 0 } }, { "id": "nightingale-farid-2022", "type": "source", "title": "AI-synthesized faces are indistinguishable from real faces and more trustworthy", "md_url": "https://research.breacher.ai/sources/nightingale-farid-2022.md", "html_url": "https://research.breacher.ai/sources/nightingale-farid-2022.html", "updated": "2026-10-07", "summary": "StyleGAN2 faces were classified at chance (48.2%); training+feedback raised accuracy only to 59.0%; synthetic faces were rated 7.7% more trustworthy than real faces.", "tags": [ "human-detection", "hyperrealism", "trust", "training" ] }, { "id": "miller-2023-ai-hyperrealism", "type": "source", "title": "AI hyperrealism: Why AI faces are perceived as more real than human ones", "md_url": "https://research.breacher.ai/sources/miller-2023-ai-hyperrealism.md", "html_url": "https://research.breacher.ai/sources/miller-2023-ai-hyperrealism.html", "updated": "2026-10-07", "summary": "White AI faces were judged human 65.9% of the time vs 51.1% for real White faces ('AI hyperrealism'); the worst detectors were the most confident; a classifier on human-rated attributes reached 94%.", "tags": [ "human-detection", "hyperrealism", "overconfidence", "uncanny-valley" ] }, { "id": "diel-2024-deepfake-detection-meta", "type": "source", "title": "Human performance in detecting deepfakes: A systematic review and meta-analysis of 56 papers", "md_url": "https://research.breacher.ai/sources/diel-2024-deepfake-detection-meta.md", "html_url": "https://research.breacher.ai/sources/diel-2024-deepfake-detection-meta.html", "updated": "2026-10-07", "summary": "Pooled human deepfake-detection accuracy was 55.54% (95% CI 48.87-62.10), not significantly above chance; detection-improvement strategies raised it to 65.14%.", "tags": [ "human-detection", "meta-analysis", "modality", "training" ] }, { "id": "kobis-2021-fooled-twice", "type": "source", "title": "Fooled twice: People cannot detect deepfakes but think they can", "md_url": "https://research.breacher.ai/sources/kobis-2021-fooled-twice.md", "html_url": "https://research.breacher.ai/sources/kobis-2021-fooled-twice.html", "updated": "2026-10-07", "summary": "People could not reliably detect deepfake videos, were biased toward judging them authentic, overestimated their ability, and neither incentives nor awareness raising helped.", "tags": [ "human-detection", "overconfidence", "response-bias", "incentives" ] }, { "id": "groh-2022-crowds-machines", "type": "source", "title": "Deepfake detection by human crowds, machines, and machine-informed crowds", "md_url": "https://research.breacher.ai/sources/groh-2022-crowds-machines.md", "html_url": "https://research.breacher.ai/sources/groh-2022-crowds-machines.html", "updated": "2026-10-07", "summary": "Ordinary human crowds and the leading DFDC model were similarly accurate but made different mistakes; model-informed participants beat either alone on average, but inaccurate model predictions often lowered their accuracy.", "tags": [ "human-detection", "human-machine-teaming", "face-processing" ] }, { "id": "mai-2023-speech-deepfakes", "type": "source", "title": "Warning: Humans cannot reliably detect speech deepfakes", "md_url": "https://research.breacher.ai/sources/mai-2023-speech-deepfakes.md", "html_url": "https://research.breacher.ai/sources/mai-2023-speech-deepfakes.html", "updated": "2026-10-07", "summary": "Listeners were 70.35% accurate on single clips (73% hit rate on fakes); familiarisation added only ≈3.8 points; crowd-aggregated judgments rivalled in-domain detectors, while out-of-domain detectors failed.", "tags": [ "human-detection", "voice", "training", "human-vs-machine" ] }, { "id": "bray-2023-deepfake-faces", "type": "source", "title": "Testing human ability to detect 'deepfake' images of human faces", "md_url": "https://research.breacher.ai/sources/bray-2023-deepfake-faces.md", "html_url": "https://research.breacher.ai/sources/bray-2023-deepfake-faces.html", "updated": "2026-10-07", "summary": "Accuracy was 62.18%, barely above chance; advice and familiarisation did not significantly help; confidence was high (6.83/10) and unrelated to accuracy.", "tags": [ "human-detection", "training", "overconfidence" ] }, { "id": "frank-2024-representative-study", "type": "source", "title": "A Representative Study on Human Detection of Artificially Generated Media Across Countries", "md_url": "https://research.breacher.ai/sources/frank-2024-representative-study.md", "html_url": "https://research.breacher.ai/sources/frank-2024-representative-study.html", "updated": "2026-10-07", "summary": "Most participants were effectively guessing; AI-generated media was rated more human-like than real media across all media types and countries.", "tags": [ "human-detection", "cross-cultural", "hyperrealism" ] }, { "id": "cooke-2024-coin-toss", "type": "source", "title": "As good as a coin toss: Human detection of AI-generated images, videos, audio, and audiovisual stimuli", "md_url": "https://research.breacher.ai/sources/cooke-2024-coin-toss.md", "html_url": "https://research.breacher.ai/sources/cooke-2024-coin-toss.html", "updated": "2026-10-07", "summary": "Mean detection was close to 50% chance; accuracy dropped for any synthetic content, foreign-language content, single-modality stimuli and synthetic images of faces; prior knowledge of synthetic media did not help; older participants did worse.", "tags": [ "human-detection", "modality" ] }, { "id": "groh-2024-political-speech", "type": "source", "title": "Human detection of political speech deepfakes across transcripts, audio, and video", "md_url": "https://research.breacher.ai/sources/groh-2024-political-speech.md", "html_url": "https://research.breacher.ai/sources/groh-2024-political-speech.html", "updated": "2026-10-07", "summary": "Accuracy rose with each added modality (transcript 57% → silent video 64% → audio 81% → video+audio 85-86% in Exp 1a); people relied more on how something was said than on what was said, and TTS fakes were harder to spot than voice-actor fakes.", "tags": [ "human-detection", "modality", "voice" ] }, { "id": "barrington-2025-voice-clones", "type": "source", "title": "People are poorly equipped to detect AI-powered voice clones", "md_url": "https://research.breacher.ai/sources/barrington-2025-voice-clones.md", "html_url": "https://research.breacher.ai/sources/barrington-2025-voice-clones.html", "updated": "2026-10-12", "summary": "Listeners judged a commercial voice clone to be the same person as the real speaker about 80% of the time (median 83.3% of real-clone pairs) and identified clones as AI only 60.8% of the time.", "tags": [ "human-detection", "voice", "voice-clone", "identity", "vishing" ] }, { "id": "gray-2025-super-recognizers", "type": "source", "title": "Training human super-recognizers' detection and discrimination of AI-generated faces", "md_url": "https://research.breacher.ai/sources/gray-2025-super-recognizers.md", "html_url": "https://research.breacher.ai/sources/gray-2025-super-recognizers.html", "updated": "2026-10-12", "summary": "Untrained controls labelled only 31% of StyleGAN3 faces as AI (below chance) and super-recognisers 41%; ≈5 minutes of artefact training raised these to 51% and 64%, with no retention test.", "tags": [ "human-detection", "training", "individual-differences", "hyperrealism" ] }, { "id": "lu-2023-fake2m", "type": "source", "title": "Seeing is not always believing: Benchmarking human and model perception of AI-generated images", "md_url": "https://research.breacher.ai/sources/lu-2023-fake2m.md", "html_url": "https://research.breacher.ai/sources/lu-2023-fake2m.html", "updated": "2026-10-07", "summary": "Humans misclassified 38.7% of images; the best model failed on 13% under the same conditions.", "tags": [ "human-detection", "text-to-image", "human-vs-machine" ] }, { "id": "lago-2022-more-real-than-real", "type": "source", "title": "More real than real: A study on human visual perception of synthetic faces", "md_url": "https://research.breacher.ai/sources/lago-2022-more-real-than-real.md", "html_url": "https://research.breacher.ai/sources/lago-2022-more-real-than-real.html", "updated": "2026-10-07", "summary": "An early report that StyleGAN2 faces were rated as more real than real faces.", "tags": [ "human-detection", "hyperrealism" ] }, { "id": "tucciarelli-2022-realness", "type": "source", "title": "On the realness of people who do not exist: The social processing of artificial faces", "md_url": "https://research.breacher.ai/sources/tucciarelli-2022-realness.md", "html_url": "https://research.breacher.ai/sources/tucciarelli-2022-realness.html", "updated": "2026-10-07", "summary": "GAN faces were judged as real as, or more real than, real faces; faces rated more trustworthy were more often judged real.", "tags": [ "human-detection", "hyperrealism", "trust" ] }, { "id": "iproov-2025-consumer-study", "type": "source", "title": "iProov deepfake blindspot consumer study (industry report)", "md_url": "https://research.breacher.ai/sources/iproov-2025-consumer-study.md", "html_url": "https://research.breacher.ai/sources/iproov-2025-consumer-study.html", "updated": "2026-10-07", "summary": "Vendor-reported: only ≈0.1% of participants classified every item correctly; older adults performed worse; confidence was high regardless of accuracy.", "tags": [ "human-detection", "overconfidence" ] }, { "id": "ivory-2026-faces-voices-domain", "type": "source", "title": "Human detection of AI-generated faces and voices is not domain-general", "md_url": "https://research.breacher.ai/sources/ivory-2026-faces-voices-domain.md", "html_url": "https://research.breacher.ai/sources/ivory-2026-faces-voices-domain.html", "updated": "2026-10-12", "summary": "Faces were detected above chance (65.48%) but cloned voices barely (53.95%); face and voice skill were not significantly related, and confidence tracked accuracy for faces but not voices.", "tags": [ "human-detection", "voice", "voice-clone", "confidence", "individual-differences" ] }, { "id": "lavan-2025-voice-clones-hyperrealism", "type": "source", "title": "Voice clones sound realistic but not (yet) hyperrealistic", "md_url": "https://research.breacher.ai/sources/lavan-2025-voice-clones-hyperrealism.md", "html_url": "https://research.breacher.ai/sources/lavan-2025-voice-clones-hyperrealism.html", "updated": "2026-10-12", "summary": "Listeners could not tell voice clones from the real speakers (d' 0.10-0.13) and rated them as real as human voices, but clones were not judged more human than humans (no hyperrealism); AI voices were rated more dominant.", "tags": [ "human-detection", "voice", "voice-clone", "hyperrealism", "trust" ] }, { "id": "pehlivanoglu-2026-machines-humans", "type": "source", "title": "Is this real? Susceptibility to deepfakes in machines and humans", "md_url": "https://research.breacher.ai/sources/pehlivanoglu-2026-machines-humans.md", "html_url": "https://research.breacher.ai/sources/pehlivanoglu-2026-machines-humans.html", "updated": "2026-10-12", "summary": "On StyleGAN2 faces humans were at chance (49%, AUC 0.53) with a strong truth bias while a CNN scored 97%; on videos the pattern reversed: machines near chance, humans better.", "tags": [ "human-detection", "human-vs-machine", "truth-bias", "confidence", "individual-differences" ] }, { "id": "dunn-2026-super-recognizers-average", "type": "source", "title": "Too good to be true: Synthetic AI faces are more average than real faces and super-recognizers know it", "md_url": "https://research.breacher.ai/sources/dunn-2026-super-recognizers-average.md", "html_url": "https://research.breacher.ai/sources/dunn-2026-super-recognizers-average.html", "updated": "2026-10-12", "summary": "Super-recognisers detected AI faces modestly above chance (57.3%, d' 0.41) while controls were at chance (50.7%); AI faces sit closer to the centre of face space, and super-recognisers use this 'hyper-averageness' as a cue.", "tags": [ "human-detection", "individual-differences", "hyperrealism", "mechanism" ] }, { "id": "dawel-2026-training-ai-faces", "type": "source", "title": "Training humans to detect AI-generated faces", "md_url": "https://research.breacher.ai/sources/dawel-2026-training-ai-faces.md", "html_url": "https://research.breacher.ai/sources/dawel-2026-training-ai-faces.html", "updated": "2026-10-12", "summary": "Training people to notice the global impressions AI faces systematically evoke (rather than artefacts) improved every participant, roughly doubled mean accuracy, and made confidence well calibrated; a test-retest control ruled out practice effects.", "tags": [ "human-detection", "training", "confidence", "hyperrealism" ] }, { "id": "rehman-2025-training-synthetic-faces", "type": "source", "title": "Training humans for synthetic face image detection", "md_url": "https://research.breacher.ai/sources/rehman-2025-training-synthetic-faces.md", "html_url": "https://research.breacher.ai/sources/rehman-2025-training-synthetic-faces.html", "updated": "2026-10-12", "summary": "One-shot artefact training raised trained participants' accuracy on synthetic faces (79.08% vs 58.29% for controls) but lowered accuracy on real faces, so overall accuracy did not differ significantly between groups.", "tags": [ "human-detection", "training", "criterion-shift" ] }, { "id": "kramer-2024-crowds-synthetic-faces", "type": "source", "title": "Crowds Improve Human Detection of AI-Synthesised Faces", "md_url": "https://research.breacher.ai/sources/kramer-2024-crowds-synthetic-faces.md", "html_url": "https://research.breacher.ai/sources/kramer-2024-crowds-synthetic-faces.html", "updated": "2026-10-12", "summary": "Aggregating different people's judgements (outer crowds) improved synthetic-face detection as crowd size grew, with gains comparable to training; averaging one person's two attempts (inner crowd) added little.", "tags": [ "human-detection", "crowds", "aggregation" ] }, { "id": "kramer-2026-aggregation-human-ai", "type": "source", "title": "Improving human performance in deepfake detection: limits of aggregation and human–AI collaboration", "md_url": "https://research.breacher.ai/sources/kramer-2026-aggregation-human-ai.md", "html_url": "https://research.breacher.ai/sources/kramer-2026-aggregation-human-ai.html", "updated": "2026-10-18", "summary": "Individuals detected deepfake videos above chance but modestly; whole-sample aggregation helped, but confidence did not discriminate correct from incorrect trials, confidence-weighted pairing reduced accuracy, and gains from accurate AI guidance were offset by following incorrect advice.", "tags": [ "human-detection", "crowds", "aggregation", "metacognition", "human-ai-teaming" ] }, { "id": "plohl-2026-political-deepfakes-plausibility", "type": "source", "title": "How deepfake plausibility, media literacy, and personal attitudes shape detection, liking, and intentions to share political deepfakes on social media", "md_url": "https://research.breacher.ai/sources/plohl-2026-political-deepfakes-plausibility.md", "html_url": "https://research.breacher.ai/sources/plohl-2026-political-deepfakes-plausibility.html", "updated": "2026-10-18", "summary": "High-plausibility political deepfakes were detected less often and liked more; higher media literacy predicted better detection, less liking and lower sharing intention; favourable attitudes to the depicted person raised liking and sharing.", "tags": [ "human-detection", "political-deepfakes", "media-literacy", "sharing" ] }, { "id": "san-segundo-2026-audio-language-style", "type": "source", "title": "Human perception of audio deepfakes: the role of language and speaking style", "md_url": "https://research.breacher.ai/sources/san-segundo-2026-audio-language-style.md", "html_url": "https://research.breacher.ai/sources/san-segundo-2026-audio-language-style.html", "updated": "2026-10-18", "summary": "Spanish and Japanese listeners averaged 59.11% when classifying natural vs synthetic speech, doing better on real voices, and relied mainly on prosodic and extralinguistic cues (intonation, rhythm, fluency, pauses, breathing, laughter).", "tags": [ "human-detection", "audio", "cross-linguistic", "cues" ] }, { "id": "manhede-2026-online-markets-training", "type": "source", "title": "Human discernment of artificial intelligence in online markets can be shaped by experience and training", "md_url": "https://research.breacher.ai/sources/manhede-2026-online-markets-training.md", "html_url": "https://research.breacher.ai/sources/manhede-2026-online-markets-training.html", "updated": "2026-10-18", "summary": "Baseline discernment of AI-generated vs human rental listings was 57.4%; heavier AI engagement predicted worse discernment (r = -0.29) while AI attitudes did not; labelled training with feedback added 9.4 points, control training none.", "tags": [ "human-detection", "ai-text", "training", "individual-differences" ] }, { "id": "sulimani-2026-recognition-ai-faces", "type": "source", "title": "Recognition of real and artificial intelligence-generated faces", "md_url": "https://research.breacher.ai/sources/sulimani-2026-recognition-ai-faces.md", "html_url": "https://research.breacher.ai/sources/sulimani-2026-recognition-ai-faces.html", "updated": "2026-10-18", "summary": "AI faces were recognised (remembered) more accurately than real faces, both face types were classified above chance with no difference, and recognition and classification accuracy were not correlated.", "tags": [ "face-recognition", "memory", "human-detection", "covert-recognition" ] }, { "id": "yeung-2026-isthisai-strategies", "type": "source", "title": "Is This AI? Longitudinal Analysis of Strategies Used for AI Detection on Two Subreddits", "md_url": "https://research.breacher.ai/sources/yeung-2026-isthisai-strategies.md", "html_url": "https://research.breacher.ai/sources/yeung-2026-isthisai-strategies.html", "updated": "2026-10-18", "summary": "Catalogued 12 strategies people use in the wild to judge whether content is AI-generated (e.g., close inspection of physical details, recognising AI content trends, assumptions about model capabilities) and showed that strategies shift over time with AI capabilities and online trends.", "tags": [ "detection-strategies", "folk-heuristics", "crowds" ] }, { "id": "mori-2012-uncanny-valley", "type": "source", "title": "The uncanny valley [From the Field]", "md_url": "https://research.breacher.ai/sources/mori-2012-uncanny-valley.md", "html_url": "https://research.breacher.ai/sources/mori-2012-uncanny-valley.html", "updated": "2026-10-07", "summary": "Proposes that affinity rises with human likeness, drops sharply just short of full likeness, then recovers; movement amplifies the curve; designers should aim for the first peak.", "tags": [ "uncanny-valley", "theory" ] }, { "id": "mathur-reichling-2016-cartography", "type": "source", "title": "Navigating a social world with robot partners: A quantitative cartography of the Uncanny Valley", "md_url": "https://research.breacher.ai/sources/mathur-reichling-2016-cartography.md", "html_url": "https://research.breacher.ai/sources/mathur-reichling-2016-cartography.html", "updated": "2026-10-07", "summary": "Demonstrated an uncanny valley in explicit likability of real robot faces that extended to implicit trust decisions; category confusion may occur but did not mediate the effect.", "tags": [ "uncanny-valley", "trust", "valley-shape" ] }, { "id": "mathur-2020-uncanny-not-confusing", "type": "source", "title": "Uncanny but not confusing: Multisite study of perceptual category confusion in the Uncanny Valley", "md_url": "https://research.breacher.ai/sources/mathur-2020-uncanny-not-confusing.md", "html_url": "https://research.breacher.ai/sources/mathur-2020-uncanny-not-confusing.html", "updated": "2026-10-07", "summary": "The likability valley replicated across sites; perceptual category confusion again did not mediate it.", "tags": [ "uncanny-valley", "replication", "valley-shape" ] }, { "id": "katsyri-2015-uv-review", "type": "source", "title": "A review of empirical evidence on different uncanny valley hypotheses: support for perceptual mismatch as one road to the valley of eeriness", "md_url": "https://research.breacher.ai/sources/katsyri-2015-uv-review.md", "html_url": "https://research.breacher.ai/sources/katsyri-2015-uv-review.html", "updated": "2026-10-07", "summary": "The simplest valley hypothesis and the movement claim lacked consistent support; categorization ambiguity had too little research to judge; perceptual mismatch received good support; the valley appears only under specific conditions.", "tags": [ "uncanny-valley", "mechanisms", "review" ] }, { "id": "diel-2022-uv-meta", "type": "source", "title": "A Meta-analysis of the Uncanny Valley's Independent and Dependent Variables", "md_url": "https://research.breacher.ai/sources/diel-2022-uv-meta.md", "html_url": "https://research.breacher.ai/sources/diel-2022-uv-meta.html", "updated": "2026-10-07", "summary": "Across 72 studies the uncanny valley effect was large (Hedges' g = 1.01); face distortion produced the largest effect (g = 1.46), followed by distinct entities, realism rendering and morphing.", "tags": [ "uncanny-valley", "meta-analysis", "mechanisms", "measurement" ] }, { "id": "macdorman-chattopadhyay-2016", "type": "source", "title": "Reducing consistency in human realism increases the uncanny valley effect; increasing category uncertainty does not", "md_url": "https://research.breacher.ai/sources/macdorman-chattopadhyay-2016.md", "html_url": "https://research.breacher.ai/sources/macdorman-chattopadhyay-2016.html", "updated": "2026-10-07", "summary": "Reducing realism consistency made humans and animals (not objects) eerier and colder; the most category-ambiguous stimuli were neither the eeriest nor the coldest, contradicting the category-uncertainty account.", "tags": [ "uncanny-valley", "mechanisms" ] }, { "id": "seyama-nagayama-2007", "type": "source", "title": "The uncanny valley: Effect of realism on the impression of artificial human faces", "md_url": "https://research.breacher.ai/sources/seyama-nagayama-2007.md", "html_url": "https://research.breacher.ai/sources/seyama-nagayama-2007.html", "updated": "2026-10-07", "summary": "Realism alone did not produce unpleasantness; strong unpleasantness appeared only when highly realistic faces had abnormal eye size.", "tags": [ "uncanny-valley", "mechanisms", "atypicality" ] }, { "id": "cheetham-2014-happy-valley", "type": "source", "title": "Perceptual discrimination difficulty and familiarity in the Uncanny Valley: more like a 'Happy Valley'", "md_url": "https://research.breacher.ai/sources/cheetham-2014-happy-valley.md", "html_url": "https://research.breacher.ai/sources/cheetham-2014-happy-valley.html", "updated": "2026-10-07", "summary": "Contrary to the uncanny valley hypothesis, category-ambiguous faces were easier (not harder) to discriminate, and greater discrimination difficulty went with more positive affect — a 'Happy Valley'.", "tags": [ "uncanny-valley", "categorical-perception" ] }, { "id": "burleigh-2013-does-uv-exist", "type": "source", "title": "Does the uncanny valley exist? An empirical test of the relationship between eeriness and the human likeness of digitally created faces", "md_url": "https://research.breacher.ai/sources/burleigh-2013-does-uv-exist.md", "html_url": "https://research.breacher.ai/sources/burleigh-2013-does-uv-exist.html", "updated": "2026-10-07", "summary": "Eeriness was mostly a linear function of human likeness; non-linear effects appeared only when stimuli combined conflicting or atypical features.", "tags": [ "uncanny-valley", "valley-shape" ] }, { "id": "steckenfinger-ghazanfar-2009", "type": "source", "title": "Monkey visual behavior falls into the uncanny valley", "md_url": "https://research.breacher.ai/sources/steckenfinger-ghazanfar-2009.md", "html_url": "https://research.breacher.ai/sources/steckenfinger-ghazanfar-2009.html", "updated": "2026-10-07", "summary": "Macaques looked less at realistic synthetic faces than at real or unrealistic ones—an uncanny-valley analogue in another species.", "tags": [ "uncanny-valley", "evolution" ] }, { "id": "wang-2015-uv-existence-explanations", "type": "source", "title": "The uncanny valley: Existence and explanations", "md_url": "https://research.breacher.ai/sources/wang-2015-uv-existence-explanations.md", "html_url": "https://research.breacher.ai/sources/wang-2015-uv-existence-explanations.html", "updated": "2026-10-07", "summary": "Proposes that humanlike agents are first processed as human; detecting non-human cues then triggers dehumanization, which produces eeriness.", "tags": [ "uncanny-valley", "mechanisms", "theory" ] }, { "id": "gray-wegner-2012-mind-perception", "type": "source", "title": "Feeling robots and human zombies: Mind perception and the uncanny valley", "md_url": "https://research.breacher.ai/sources/gray-wegner-2012-mind-perception.md", "html_url": "https://research.breacher.ai/sources/gray-wegner-2012-mind-perception.html", "updated": "2026-10-07", "summary": "Unease was driven by attributing the capacity to feel (experience) to machines—or denying it to humans—not by appearance alone.", "tags": [ "uncanny-valley", "mind-perception" ] }, { "id": "stein-ohler-2017-uncanny-mind", "type": "source", "title": "Venturing into the uncanny valley of mind—The influence of mind attribution on the acceptance of human-like characters in a virtual reality setting", "md_url": "https://research.breacher.ai/sources/stein-ohler-2017-uncanny-mind.md", "html_url": "https://research.breacher.ai/sources/stein-ohler-2017-uncanny-mind.html", "updated": "2026-10-07", "summary": "With visuals held constant, eeriness was highest when empathic behaviour was attributed to an autonomous AI—an 'uncanny valley of mind'.", "tags": [ "uncanny-valley", "mind-perception" ] }, { "id": "saygin-2012-predictive-coding", "type": "source", "title": "The thing that should not be: predictive coding and the uncanny valley in perceiving human and humanoid robot actions", "md_url": "https://research.breacher.ai/sources/saygin-2012-predictive-coding.md", "html_url": "https://research.breacher.ai/sources/saygin-2012-predictive-coding.html", "updated": "2026-10-07", "summary": "Repetition suppression in bilateral anterior intraparietal sulcus was stronger for the android (appearance–motion mismatch) than for the human or robot, interpreted as prediction error; authors state this cannot establish a causal link to the uncanny valley.", "tags": [ "uncanny-valley", "neural", "predictive-coding" ] }, { "id": "rosenthal-von-der-putten-2019-vmpfc", "type": "source", "title": "Neural mechanisms for accepting and rejecting artificial social partners in the uncanny valley", "md_url": "https://research.breacher.ai/sources/rosenthal-von-der-putten-2019-vmpfc.md", "html_url": "https://research.breacher.ai/sources/rosenthal-von-der-putten-2019-vmpfc.html", "updated": "2026-10-07", "summary": "vmPFC encoded likability as a non-linear function of humanlikeness with selectively low likability for highly humanlike agents; behavioural valley in 18 of 21 participants; DMPFC encoded choices about artificial agents and an amygdala signal predicted their rejection.", "tags": [ "uncanny-valley", "neural", "decision-making" ] }, { "id": "schindler-2017-erp-realism", "type": "source", "title": "Differential effects of face-realism and emotion on event-related brain potentials and their implications for the uncanny valley theory", "md_url": "https://research.breacher.ai/sources/schindler-2017-erp-realism.md", "html_url": "https://research.breacher.ai/sources/schindler-2017-erp-realism.html", "updated": "2026-10-07", "summary": "N170 was U-shaped across realism (strongest for most abstract and most realistic faces), LPP increased linearly with realism, and EPN was unaffected by realism; no uncanny 'dip' was directly observed.", "tags": [ "uncanny-valley", "neural" ] }, { "id": "moshel-2022-decoding-ai-faces", "type": "source", "title": "Are you for real? Decoding realistic AI-generated faces from neural activity", "md_url": "https://research.breacher.ai/sources/moshel-2022-decoding-ai-faces.md", "html_url": "https://research.breacher.ai/sources/moshel-2022-decoding-ai-faces.html", "updated": "2026-10-07", "summary": "Group-level behavioural judgments were near chance (people often called real faces fake and vice versa), yet neural activity reliably distinguished AI-generated from real faces.", "tags": [ "uncanny-valley", "neural", "implicit-detection", "human-detection" ] }, { "id": "eiserbeck-2023-deepfake-smiles", "type": "source", "title": "Deepfake smiles matter less—the psychological and neural impact of presumed AI-generated faces", "md_url": "https://research.breacher.ai/sources/eiserbeck-2023-deepfake-smiles.md", "html_url": "https://research.breacher.ai/sources/eiserbeck-2023-deepfake-smiles.html", "updated": "2026-10-07", "summary": "Simply labelling a real smiling face as 'fake' made it rated less positive and removed its early neural smile effects (P1/N170/EPN), while increasing late, effortful processing (LPP); angry faces were unaffected by the label.", "tags": [ "uncanny-valley", "neural", "labelling", "emotion" ] }, { "id": "weisman-pena-2021-doppelganger", "type": "source", "title": "Face the uncanny: The effects of doppelganger talking head avatars on affect-based trust toward artificial intelligence technology are mediated by uncanny valley perceptions", "md_url": "https://research.breacher.ai/sources/weisman-pena-2021-doppelganger.md", "html_url": "https://research.breacher.ai/sources/weisman-pena-2021-doppelganger.html", "updated": "2026-10-07", "summary": "Self-resembling deepfake avatars increased uncanniness, which in turn reduced affect-based trust.", "tags": [ "uncanny-valley", "deepfake-video", "trust" ] }, { "id": "lee-2026-ai-chest-radiographs", "type": "source", "title": "Complementary Error Profiles of Human Readers and Large Language Models in Detecting AI-Generated Chest Radiographs: A Comparative Reader Study", "md_url": "https://research.breacher.ai/sources/lee-2026-ai-chest-radiographs.md", "html_url": "https://research.breacher.ai/sources/lee-2026-ai-chest-radiographs.html", "updated": "2026-10-19", "summary": "Expert and non-expert physicians and frontier LLMs could not reliably tell AI-generated chest X-rays from real ones: the best human (radiologist, 75.8% balanced accuracy) and best LLM (77.9%) were similar but made different errors, radiologists called 38-45% of real images fake, and a strong frequency-domain detector collapsed on a held-out generator (12% sensitivity).", "tags": [ "human-detection", "human-vs-machine", "expertise" ] }, { "id": "chen-2026-humans-ai-deepfake-teaming", "type": "source", "title": "How Humans and AI Detect Deepfakes Differently: Complementary Strategies for Human-AI Teaming", "md_url": "https://research.breacher.ai/sources/chen-2026-humans-ai-deepfake-teaming.md", "html_url": "https://research.breacher.ai/sources/chen-2026-humans-ai-deepfake-teaming.html", "updated": "2026-10-19", "summary": "Humans and AI detectors relied on different cues; eye-tracking and EEG suggested holistic, flexible human viewing that catches some inconsistencies AI misses, and people with more AI exposure matched or exceeded AI detectors in some contexts (abstract; no numbers reported).", "tags": [ "human-detection", "human-ai-teaming", "neural" ] }, { "id": "naseem-2026-in-the-wild-detection-review", "type": "source", "title": "From curated datasets to social-media reality: In-the-wild, multimodal, and generalizable deepfake detection", "md_url": "https://research.breacher.ai/sources/naseem-2026-in-the-wild-detection-review.md", "html_url": "https://research.breacher.ai/sources/naseem-2026-in-the-wild-detection-review.html", "updated": "2026-10-19", "summary": "Review concluding that strong benchmark performance of deepfake detectors does not necessarily carry over to real-world media, because of platform compression, reposting, post-processing, modality mismatch, temporal drift and new generators; calls for deployment-oriented, continually re-benchmarked detection.", "tags": [ "detector-generalization", "review" ] }, { "id": "mamsheli-2026-eeg-ai-faces", "type": "source", "title": "Processing of Real and AI-Generated Human Faces: An Exploratory EEG Study", "md_url": "https://research.breacher.ai/sources/mamsheli-2026-eeg-ai-faces.md", "html_url": "https://research.breacher.ai/sources/mamsheli-2026-eeg-ai-faces.html", "updated": "2026-10-19", "summary": "Real and AI-generated faces evoked different early brain responses (N170, N400, EPN) but not a different late LPP, and were rated differently for expressiveness and 'sense of self', while perceived intelligence and appeal did not differ.", "tags": [ "uncanny-valley", "neural", "implicit-detection" ] }, { "id": "hsieh-2026-robot-gaze-uncanny", "type": "source", "title": "Robot gaze-timing combinations and the uncanny valley", "md_url": "https://research.breacher.ai/sources/hsieh-2026-robot-gaze-uncanny.md", "html_url": "https://research.breacher.ai/sources/hsieh-2026-robot-gaze-uncanny.html", "updated": "2026-10-19", "summary": "Across nine robot gaze-timing conditions, likability followed an uncanny-valley (cubic) curve over perceived anthropomorphism while perceived empathy rose steadily; extraversion, negative attitudes and neuroticism moderated these relations.", "tags": [ "uncanny-valley", "valley-shape", "individual-differences" ] }, { "id": "jang-2026-uncanny-conversation", "type": "source", "title": "Uncanny Valley in Conversation? Effects of Internal and External Human-Likeness on Intimacy and Trust", "md_url": "https://research.breacher.ai/sources/jang-2026-uncanny-conversation.md", "html_url": "https://research.breacher.ai/sources/jang-2026-uncanny-conversation.html", "updated": "2026-10-19", "summary": "A machine-like AI counsellor was liked and trusted most at low self-disclosure and less at high disclosure (an uncanny-valley-like curve of mind), whereas a human-like counsellor gained with deeper disclosure; internal and external human-likeness interact.", "tags": [ "uncanny-valley", "mind-perception", "trust" ] } ] }