Confident Security
Confident Security builds CONFSEC, a confidential-computing layer that lets enterprises use third-party AI models while cryptographically proving prompts are never logged or used for training.
Visit Website ↗ + Add to Compare Claim This CompanyOverview
Confident Security, founded in San Francisco in 2024 by Jonathan Mortensen (a two-time founder with prior exits to BlueVoyant and Databricks), addresses a specific enterprise AI adoption blocker: regulated organizations often cannot use third-party LLM APIs because they cannot verify that sensitive prompts and outputs stay private.
Its CONFSEC platform sits between AI vendors and enterprise customers as a verifiable privacy layer, built on OpenPCC — an open standard the company created, modeled conceptually on Apple’s Private Cloud Compute architecture. It combines trusted execution environments, blind signatures, and remote attestation so that an enterprise, or an independent auditor, can cryptographically verify that prompts were never logged, retained, or used for model training, rather than simply trusting a vendor’s privacy policy.
The company raised $4.2 million in seed funding in 2025 from Decibel, South Park Commons, Ex Ante, and Swyx, and its team draws from Google, Apple, Databricks, Red Hat, and HashiCorp. It targets finance, healthcare, and government customers, though as a young, seed-stage company it has not yet published named enterprise deployments.
Innovation Matrix Assessment
Created an open standard (OpenPCC) and shipped a working confidential-computing layer for third-party AI inference within roughly a year of founding.
Removes a real adoption blocker for regulated enterprises that want to use third-party LLMs but cannot risk exposing sensitive prompts and data.
Only $4.2M in seed funding and no publicly named customers or pilots; very early commercially despite a strong founding team.
Applying Apple Private Cloud Compute-style verifiable confidential computing specifically as a third-party AI inference wrapper is a distinctive approach few competitors offer.
The architecture is technically sound in principle and open-source/auditable, but no independent verification or named production use was found yet.
As enterprises adopt more third-party AI and LLM services, privacy-preserving inference layers address a durable and growing compliance need.
Why CISOs Should Care
Lets regulated enterprises adopt third-party AI models while providing cryptographic, auditable proof that sensitive prompts are never logged or used for training, rather than relying on a vendor's word.
What Makes It Different
Built an open standard (OpenPCC) for verifiable AI inference privacy, using remote attestation so privacy claims can be independently checked rather than merely asserted.
The Matrix Verdict
53/100 — INCREMENTAL INNOVATOR
An Incremental Innovator: a technically strong, well-pedigreed team with a genuinely differentiated architecture, but still pre-revenue with no independently verifiable deployments.
Editorial Note: Claims vs. Verified Findings
The zero-logging and non-training claims are architectural and cryptographically verifiable in principle via the open OpenPCC standard, but no independent third-party audit of a live deployment was found.
Sources
Alternatives to Confident Security
Adaptive Security
AI-driven platform that simulates deepfake, voice, and multichannel social-engineering attacks to train and test organizations against next-generation phishing.
Quilr
Early-stage agentic AI security startup building a 'Service-as-Software' platform to guard against human-related breaches and secure AI agent…
Zenity
Governance and security platform for AI agents and low-code/no-code development, securing agent identity, permissions and behavior across the…
Tenzai
An agentic AI penetration testing startup building autonomous 'AI hackers' to find and validate exploitable vulnerabilities at a…
Reco
Reco secures the "agentic ecosystem" — mapping what AI agents can access across SaaS and enterprise apps, detecting…
Charm Security
Agentic AI workforce that investigates and intervenes on scams and fraud in real time, reading manipulation and intent…