We said SKIP on October 21, 2026. Not settled — due October 21, 2027.
Read this with the caveat. We tested the engine that produced this verdict against 292 launches whose outcomes we already knew, and could not show it predicted which survived. Some of its data sources were also dead at the time of scoring. The verdict stays up, dated and unedited, because a record you can quietly revise is not a record — but it is worth less than it looked when it was written.
Across all channels the pull is thin: one tangential consumer Reddit thread, a 4-point HN post, zero keyword demand, and no verified product actually charging for AI chat leak prevention. The gap list is plausible but describes an enterprise security procurement market — audits, SIEM integrations, central policy engines — which is the worst possible first market for a solo founder with no audience. SKIP unless the operator already has security/compliance domain credibility and a warm channel to CISOs.
The only genuine signal is one consumer thread about Character.AI chats leaking — anxiety, not purchase intent — plus a Show HN at 4 points, which is effectively dead on arrival. The three 'competitors' listed are not actually paying DLP products: two are LLM vendors where privacy is a feature, so there is no proof anyone is charging money for this.
A wedge exists on paper (browser-extension redaction before submission to public LLMs), but every cited gap — HIPAA/PCI-DSS/SOC2 compliance tooling, SIEM/Purview integrations, central policy enforcement for shadow AI — describes entrenched enterprise security procurement, which a solo founder cannot reach without audits, references and a sales motion.
The strongest case for building: 'shadow AI' is a real and growing security headache, and a dumb-simple browser extension that redacts API keys, PII and credentials before you hit Enter is a weekend build with no procurement required — you can sell it bottom-up to individual developers and small teams at $9/seat and let it creep into companies the way Grammarly did. If even one regulated-industry customer bites, the compliance story writes itself and the moat compounds through policy libraries.
What is worth more than this page. The register records what became of 14,962 real launches. No engine has to be right for that to be true.