Secludy
San Francisco startup generating differentially-private synthetic data so regulated companies can train AI models without exposing real customer PII.
Visit Website ↗ + Add to Compare Claim This CompanyOverview
Secludy generates privacy-guaranteed synthetic datasets that mirror a customer’s real data so banks, payments firms and other regulated companies can train, fine-tune and evaluate AI models without exposing actual customer PII. The platform is built on differential privacy techniques, is designed to be compliant by design with GDPR, CCPA and HIPAA, and deploys entirely within the customer’s own cloud environment rather than sending data to Secludy’s infrastructure.
Founded in 2024 in San Francisco by CEO Ben Cerchio and Ming He, Secludy publicly launched its platform and closed a $4 million seed round in May 2026, led by fintech-focused Impression Ventures with participation from LAUNCH, The Syndicate, Wedbush Ventures, Precursor Ventures, Hustle Fund, Script Capital, Mana Ventures and Chispa VC. The company’s stated differentiator is treating data masking and traditional anonymization as insufficient for large language models, arguing that only differential-privacy-based synthetic generation reliably prevents PII leakage during AI training.
Innovation Matrix Assessment
Moved from founding to a public platform launch and $4M seed within about two years, though the fundraise and launch happened simultaneously, suggesting an early product.
Addresses a real and growing problem — regulated firms wanting to use proprietary data for GenAI without PII exposure — but adoption evidence is limited to the funding announcement.
Single $4M seed round, no disclosed named enterprise customers or revenue figures found.
Differential-privacy synthetic data is an established academic and commercial technique (competitors include established DSPM and synthetic-data vendors); Secludy's focus on financial-services GenAI training is a niche application rather than a new category.
No independent benchmarking, named case studies, or third-party validation of privacy guarantees or synthetic data quality were found.
As regulated industries increasingly want to train AI on proprietary data, privacy-preserving data generation is likely to remain a persistent need over the next several years.
Why CISOs Should Care
Lets data-protection and privacy teams unlock proprietary data for AI initiatives without the compliance risk of exposing real PII to model training pipelines.
What Makes It Different
Built specifically on differential privacy (a mathematically defined privacy guarantee) rather than heuristic masking or tokenization, and deploys inside the customer's own cloud rather than a shared vendor environment.
The Matrix Verdict
37/100 — EMERGING / UNRANKED
Emerging tier: a real, credibly funded company addressing a specific and growing compliance need, but with no independent efficacy evidence or disclosed customers, it is too early to place higher.
Editorial Note: Claims vs. Verified Findings
Funding, investors and founder identities are independently confirmed by GlobeNewswire and HPCwire/BigDATAwire coverage; claims about privacy guarantees and synthetic data quality are the company's own and unverified independently.
Sources
- GlobeNewswire — https://www.globenewswire.com/news-release/2026/05/13/3294051/0/en/secludy-raises-4m-to-safely-unlock-proprietary-data-for-genai.html
- BigDATAwire/HPCwire — https://www.hpcwire.com/bigdatawire/this-just-in/secludy-raises-4m-to-safely-unlock-proprietary-data-for-genai/
- Secludy — https://secludy.com/
Alternatives to Secludy
Fireblocks
Fireblocks is the category-defining institutional digital-asset infrastructure provider, with roughly $2 trillion transferred and the industry's highest valuation…
Cyera
AI-powered, agentless data security platform combining DSPM, DLP, and AI-activity monitoring, and one of the fastest-funded startups in…
Varonis
Data-centric security platform that monitors file, email, and cloud activity to detect insider threats and ransomware before data…
Bedrock Data
AI-native data security posture management (DSPM) platform using a Metadata Lake to discover, classify, and contextualize data across…
Halcyon
Ransomware-focused cyber resilience platform combining endpoint prevention, automated recovery, and data exfiltration protection.
CyberRidge
Photonic-layer encryption that converts transmitted data into optical noise, defending fiber communications against quantum-era decryption.