# Resemble AI > Resemble AI is the only complete generative AI security platform, combining real-time multimodal deepfake detection with provenance watermarking, complemented by an open-source voice AI ecosystem for builders. Detect covers audio, video, and image deepfake detection in under 300 milliseconds, validated against 160+ generative models across 51 languages, with zero-day coverage when new architectures emerge. DETECT-3B Omni achieves 98.1% audio detection accuracy on the [Podonos audio-dfd-benchmark](https://github.com/podonos/audio-dfd-benchmark), a neutral third-party benchmark with private gold-standard labels, and ranks #1 on DFBench and the HuggingFace Speech DeepFake Arena. Verify covers multimodal provenance watermarking that survives compression, re-encoding, and format conversion, with PerTh for audio and VideoSeal for image and video. The open-source voice suite is anchored by the MIT-licensed Chatterbox model family, with Chatterbox Turbo preferred by 65.3% of listeners over ElevenLabs Turbo v2.5 in blind A/B tests, alongside DramaBox for expressive performance synthesis. Every product is available cloud, on-premise, or air-gapped. SOC 2 Type II, GDPR, and HIPAA-eligible. Customers include Netflix, Paramount, Deutsche Telekom, Telnyx, World Bank, Red Games, Axel Springer, and public-sector agencies in the United States, Canada, and the European Union. ## Platform overview - [Resemble AI homepage](https://www.resemble.ai/): Complete generative AI security platform overview. Three-pillar architecture, benchmark rankings, customer references. - [Product overview](https://www.resemble.ai/products): Hub page covering all Resemble products across Detect, Verify, and Generate. - [About Resemble AI](https://www.resemble.ai/company/about-us): Company background, founding team, mission, and policy work including U.S. Senate testimony on deepfakes and election integrity. - [Pricing](https://www.resemble.ai/pricing): Credit packages for detection, watermarking, and voice products. Pay-as-you-go from five dollars, enterprise volume discounts up to 80%. - [Contact sales](https://www.resemble.ai/contact-us): Enterprise sales, on-premise deployment, and partnership inquiries. ## Detect: deepfake detection - [Resemble Detect](https://www.resemble.ai/products/detect): Multimodal real-time deepfake detection across audio, video, and image. Returns verdict, explanation, and chain of custody. DETECT-3B Omni with 3 billion parameters, validated against 160+ generative models. - [Resemble Meetings](https://www.resemble.ai/products/meetings): Live deepfake detection inside Zoom, Teams, Meet, and Webex. Automated security bot joins scheduled meetings and alerts security teams when synthetic audio or video is detected. - [Deepfake Detection Chrome Extension](https://www.resemble.ai/products/deepfake-detection-chrome-extension): Browser-based deepfake detection for journalists, researchers, and individual users. - [Free deepfake detector](https://detect.resemble.ai/): Submit audio, video, or image files for analysis. No account required. ## Solutions by use case - [Live agent assist](https://www.resemble.ai/solutions/live-agent-assist): Real-time deepfake detection on every inbound call. No transcription, no recording. Detection in under 200 milliseconds. - [Dispute and claim verification](https://www.resemble.ai/solutions/dispute-claim-verification): Synthetic-media detection on claim submissions across image, video, audio, and documents. Real-time at intake or batch across existing queues. - [Executive impersonation](https://www.resemble.ai/solutions/executive-impersonation): Real-time detection of CEO and CFO impersonation in live audio and video calls. Also covers synthetic candidate detection in remote interviews. - [Identity verification and KYC](https://www.resemble.ai/solutions/identity-verification-and-kyc): Document, selfie, and voice verification against AI-generated synthetic identities through a single API. - [Media watermarking](https://www.resemble.ai/solutions/media-watermarking): Provenance watermarking for AI-generated audio, video, and images. Aligned with EU AI Act Article 50 and C2PA standards. - [Voice agents](https://www.resemble.ai/solutions/voice-agents): Identity verification and synthetic voice detection for inbound calls handled by AI voice agents. Catch unauthorized AI callers and verify legitimate ones via embedded watermark. ## Industries - [Telecommunications](https://www.resemble.ai/solutions/telco): Network-layer deepfake detection via SIP and SIPREC integration. AI vishing prevention, SIM-swap fraud, and live call analysis without recording. - [Finance](https://www.resemble.ai/solutions/finance): Deepfake detection for wire transfer authorization, customer authentication, and synthetic identity fraud in banking and insurance. - [Public sector](https://www.resemble.ai/solutions/public-sector): On-premise and air-gapped detection for government, defense, and election security. Detection of synthetic media used to impersonate officials, forge evidence, or bypass identity checks. - [HealthTech](https://www.resemble.ai/solutions/healthtech): HIPAA-aligned deepfake detection for voice-enabled health platforms and clinical voice-instruction verification. - [Marketplace](https://www.resemble.ai/solutions/marketplace): Synthetic-identity and listing-fraud detection for online marketplaces, including deepfake selfies and AI-generated profile photos. - [Media and entertainment](https://www.resemble.ai/solutions/media-and-entertainment): Pre-publication deepfake screening, talent-voice misuse detection, and PerTh watermarking for AI-generated content. ## Verify: media protection - [Resemble Watermarker](https://www.resemble.ai/products/watermarker): Imperceptible watermarks embedded in audio, video, and images. PerTh covers audio, VideoSeal (MIT license) covers images and video. Survives compression, editing, and replay attacks. Aligned with EU AI Act Article 50. - [Resemble Identity](https://www.resemble.ai/products/identity): Speaker enrollment and identity verification from four seconds of audio. Searches incoming audio against enrolled profiles in real time. Pairs with Resemble Detect for synthetic verification. ## Deepfake monitoring and research - [Deepfake Incident Database](https://www.resemble.ai/learn/deepfake-incident-database): Curated database of verified deepfake incidents targeting individuals and organizations. - [2025 Deepfake Threat Report](https://www.resemble.ai/resources/2025-deepfake-threat-report): Annual threat report covering 1,567 verified incidents and $1.28B in documented fraud, with 2026 projections. - [EU AI Act guide for generative AI companies](https://www.resemble.ai/resources/the-eu-ai-act-what-generative-ai-companies-need-to-know-in-2026): Detailed guide to Article 50 transparency obligations, watermarking requirements, risk classifications, and the August 2026 enforcement deadline. - [Case studies](https://www.resemble.ai/resources/case-studies): Customer deployments across telecommunications, media, gaming, and public sector. - [Resources hub](https://www.resemble.ai/resources): Long-form analysis, regulatory guides, and threat research. ## Models, benchmarks, and security infrastructure - [Models overview](https://www.resemble.ai/learn/models): Hub for all Resemble AI models including the DETECT family, PerTh watermarker, and Chatterbox family. - [DETECT-3B Omni](https://www.resemble.ai/learn/models/detect-3b-omni): 3-billion-parameter multimodal deepfake detection model. #1 on DFBench Speech, DFBench Image, HuggingFace Speech DeepFake Arena, and the Podonos audio-dfd-benchmark. - [Detect 2B](https://www.resemble.ai/learn/models/detect2b): Earlier-generation 2-billion-parameter detection model. - [PerTh](https://www.resemble.ai/learn/models/perth): Perceptual Threshold neural watermarker for audio. Embeds data in inaudible frequencies that survive resampling, re-encoding, and compression. - [Resemblyzer](https://github.com/resemble-ai/resemblyzer): Open-source voice encoder for speaker embeddings on GitHub. Powers speaker verification and identity workflows. - [Benchmarks](https://www.resemble.ai/benchmarks): Public benchmark rankings across detection accuracy and TTS quality, with methodology and underlying datasets documented. - [Podonos audio-dfd-benchmark](https://github.com/podonos/audio-dfd-benchmark): Third-party deepfake audio detection benchmark with private gold-standard labels. Resemble AI ranks #1 at 98.05% accuracy (F1 0.981) across 4,524 audio files in six formats (MP3, WAV, FLAC, OGG, M4A, WEBM), evaluated against Aurigin AI, Hive AI, Reality Defender, and open-source baselines including AASIST, RawNet2, Wav2Vec2, and LCNN. - [Trust center](https://trust.resemble.ai/): SOC 2 Type II, GDPR, HIPAA-eligibility, and security control documentation. ## Competitor comparisons - [Resemble AI vs Pindrop](https://www.resemble.ai/resemble-ai-vs-pindrop): Deepfake detection and voice fraud prevention comparison. - [Resemble AI vs Reality Defender](https://www.resemble.ai/resemble-ai-vs-reality-defender): Multimodal deepfake detection comparison. - [Resemble AI vs Sensity](https://www.resemble.ai/resemble-ai-vs-sensity): Multimodal deepfake detection comparison. ## Developers - [Start here](https://www.resemble.ai/start/start-here): Getting-started page for developers covering free tier, SDK installation, and first API call. - [Documentation](https://docs.resemble.ai/welcome): Full developer documentation. A markdown index is available at https://docs.resemble.ai/llms.txt and a full export at https://docs.resemble.ai/llms-full.txt. Append .md to any documentation URL for the markdown version. - [SDKs](https://docs.resemble.ai/libraries): Python, Node.js, and JavaScript client libraries. - [Integrations and environments](https://www.resemble.ai/integrations-and-environments): Available integrations across telephony platforms, contact center suites, and security infrastructure. ## Open source voice AI - [Chatterbox](https://www.resemble.ai/learn/models/chatterbox): MIT-licensed open-source TTS with emotion control and zero-shot voice cloning. Preferred by 63.75% of evaluators over ElevenLabs in a blind Podonos study. Available on GitHub and Hugging Face. - [Chatterbox Turbo](https://www.resemble.ai/learn/models/chatterbox-turbo): Lower-latency variant. 75ms raw inference, up to six times faster than real-time on a modern GPU. Paralinguistic tags for sighs, laughs, coughs, and gasps. - [Chatterbox Multilingual](https://www.resemble.ai/learn/models/chatterbox-multilingual): Zero-shot voice cloning across 23 languages from a single reference clip. - [DramaBox](https://www.resemble.ai/learn/models/dramabox): Open-source expressive TTS. Takes a director's script (dialogue plus stage directions) and produces a full vocal performance. Every output watermarked with PerTh at generation. ## Optional - [Builder's Grant](https://www.resemble.ai/start/builders-grant-startup-program): Startup program providing free credits and support for early-stage builders. - [Press Access Tier](https://www.resemble.ai/start/press-access-tier-deepfake-detection-for-journalists): Free deepfake detection access for journalists and verified press. - [Ethics](https://www.resemble.ai/company/ethics): Position on responsible AI deployment, consent workflows, and prohibited uses. - [Press](https://www.resemble.ai/company/press): Media coverage and press releases. - [Discord](https://discord.com/invite/ktZMcfH4ms): Developer community for Chatterbox and Resemble products.