About DeepSwapAI

What DeepSwapAI Is

DeepSwapAI (deepswapai.com) is a professional AI face-morphing platform serving creative professionals, marketing agencies, post-production houses, and enterprise media teams. Since launch in 2023, the platform has processed over 25 million face-morph operations through both web interface and REST API.

We are independent of the unrelated deepswap.ai product. Our consumer sister brand for casual social-media use is FaceSwapAI. See /trust for full brand disambiguation.

Technology Stack

DeepSwapAI is engineered for production workflows where output fidelity, predictable latency, and audit trails matter:

• Frontend: Next.js 15 (App Router, React 19), next-intl 4.3 for 11-language i18n, TailwindCSS 4. 6,978 routes pre-rendered as static HTML.

• Face Detection: HRNet 128-point landmark model with sub-pixel accuracy. RetinaFace multi-face detection supporting up to 50 simultaneous faces in group photos.

• Generation Pipeline: Proprietary GAN architecture for face-swap inference, with adaptive color harmonization in CIE LAB color space, automatic skin-tone matching, and 3D facial geometry reconstruction for natural perspective preservation.

• Video Pipeline: Temporal coherence engine with independent per-frame motion-vector tracking, ensuring flicker-free output. 4K (3840×2160) output on Pro tier. H.264, H.265, VP9, AV1 codecs supported.

• Infrastructure: NVIDIA A100/H100 GPU clusters, Cloudflare global CDN, BullMQ task queue, Redis 7, Prisma 6.19 ORM, PostgreSQL 15. Sub-second inference latency on Enterprise tier.

API & Workflows

Beyond the web UI, DeepSwapAI exposes a REST API at https://deepswapai.com/doc for programmatic and batch workflows:

• Authentication: Bearer token issued in account dashboard • Endpoints: 50+ endpoints covering image swap, video swap, batch operations, character animation, lip sync, voice generation, and result polling • Rate limits: 100 req/month (Starter), 1K (Popular), 5K (Pro), custom for Enterprise (typically 100K+/month) • Webhooks: For long-running operations, callbacks fire on completion with full result metadata • SDKs: Python, Node.js, TypeScript published; OpenAPI 3.1 spec available • Postman collection: Importable from /doc

Enterprise customers also get S3-compatible direct upload (bypassing web upload size limits) and dedicated GPU instances with sub-second latency SLAs.

Standards & Compliance

DeepSwapAI maintains active compliance with the following frameworks:

• GDPR (EU 2016/679, effective 2018-05-25) • CCPA (California, effective 2020-01-01) • BIPA (Illinois 740 ILCS 14) • TAKE IT DOWN Act 2025 (U.S. federal) • EU AI Act (Regulation 2024/1689, deepfake disclosure under Article 50, in force 2026-08-02) • C2PA Content Credentials (cross-industry content provenance, version 2.1) • NCMEC CyberTipline reporting protocols

For enterprise customers: SOC 2 Type II report and ISO 27001 documentation available under NDA. DPA signature available for GDPR Article 28 sub-processor agreements.

Privacy & Data Handling

Privacy is architectural, not optional:

• Auto-deletion: 24-hour standard tier, 1-hour Enterprise tier. No backups, archives, or copies retained beyond this window.

• No training on user data: Your uploads are never incorporated into model training, fine-tuning, or any algorithmic improvement. Your likeness remains exclusively yours.

• Encryption: TLS 1.3 in transit, AES-256 at rest, encrypted database columns for any account data.

• Isolated processing: Each request runs in a containerized environment with no cross-tenant access. Enterprise customers can request 'no-trace' mode where infrastructure logs exclude file metadata.

• Audit logs: Every operation is timestamped and attributable. Customers can export their own activity logs from the dashboard. Internal access is logged and audited.

• Sub-processors: Public list at /privacy. Currently: Cloudflare (CDN/DDoS), AWS (storage), GPU compute partners (regional), all under DPA.

Open Source We Build On

DeepSwapAI thanks the open-source community for the following components:

• Next.js (Vercel, MIT) — full-stack framework • next-intl (Yannick Lendrick, MIT) — i18n • React (Meta, MIT) — UI library • TailwindCSS (Tailwind Labs, MIT) — utility CSS • Prisma (Prisma Labs, Apache 2.0) — ORM • BullMQ (Manuel Lemos, MIT) — task queue • PyTorch (Meta, BSD) — deep learning framework foundation • CUDA (NVIDIA) — GPU computation • Schema.org — structured data vocabulary

We contribute back where applicable through bug reports and documentation patches.

Contact

For:

• Sales / partnerships: [email protected] • General support: [email protected] • Privacy / GDPR / DPO: [email protected] • Safety / abuse: [email protected] • Copyright / DMCA: [email protected] • Security disclosures (RFC 9116): [email protected] or /.well-known/security.txt • Legal / regulatory: [email protected]

Response SLA: Sales 4h. General support 24h business hours. Safety 4h. Critical security 1h.

For casual / consumer face-swap use (free tier, social-media-first), see our sister platform at https://faceswapai.com.