01
Executive summary
PurityOS is an operating system with crime prevention built in. Local AI on the device and cloud AI that tracks threats across the internet work together to stop exploitation, pornography, scams and fraud, drug and alcohol sales, dangerous weapons and violent extremism, digital piracy, cyberbullying, harassment and sextortion, misinformation, insider data theft and malware — before they reach the screen. Every user is identity-verified once at setup. PurityOS ships in two editions: installed on the device, or PurityOS Cloud, streamed to any screen. It is developed by AW3 Technology Venture Studio.
21.3M
CyberTipline reports to NCMEC in 2025; online enticement up 156% (NCMEC)
$20.9B
cyber-enabled losses reported to the FBI in 2025, up 26% (IC3 2025)
1 Jan 2027
California AB 1043 places age-assurance duties directly on operating system providers (Hunton)
Why now. For thirty years, US law pointed prevention at the user's device — Section 230 states a policy of encouraging "blocking and filtering technologies that empower parents" (47 U.S.C. § 230) — but no one built that layer as a default-on, tamper-resistant product. Regulators are now making the OS the place for safety: California AB 1043, the UK's September 2026 commitment to legislate device-level protections (Guardian), and Australia's codes binding OS vendors since March 2026 (eSafety). PurityOS is the product the 1996 policy framework assumed would exist.
The business. Subscriptions for individuals ($6/mo) and families ($12/mo), plus per-seat contracts for schools and employers. We enter where the buyer and the user differ — guardians, schools, faith communities and employers — and distribute through school-to-home conversion, church networks and OEM partners. An illustrative obtainable scenario reaches roughly $30M ARR in the US alone (see Market).
What must be true. Tamper resistance on partner hardware, independently audited classifier accuracy, a privacy-preserving identity design that stores attestations rather than documents, and a build on existing OS foundations (AOSP / ChromiumOS) rather than from scratch.
02
Mission
Removing crime from the digital world. Our mission is to remove crime from the digital world — starting where every digital crime begins, the operating system — by preventing crime and cybercrime, from exploitation and fraud to data theft and malware, for everyone who chooses a safer digital life and everyone entrusted to someone else’s care.
Why the operating system
- Every digital crime passes through an OS. Grooming, scams, drug and weapon sales and data theft arrive through different apps, but all of them run on an operating system — the one layer where protection covers everything at once.
- Protection on top can be taken away. Apps, extensions and network filters can be closed, bypassed or uninstalled. Protection inside the operating system can’t be removed by the software it guards against.
- The internet was built to deliver, not to discern. The network moves data and leaves judgement to the device. US law pointed the same way, encouraging blocking technology that empowers parents — but devices never took that on by default.
- Prevention protects; prosecution responds. Handling crime after the fact means there is already a victim. Stopping harm before it reaches the screen means it never happens.
- Safety and privacy belong together. Local AI analyses personal content on the device; cloud AI tracks threats without seeing anyone’s messages, photos or history.
- Accountability ends anonymous abuse. Every user is identity-verified once — children through a guardian — so no one hides behind a throwaway identity or poses as a child.
- The law is moving to the OS. California’s Digital Age Assurance Act (AB 1043), in force from 1 January 2027, places age-assurance duties directly on operating system providers.
03
The problem
Reported online harm reached record levels in 2025, and it grows fastest where it crosses apps: harm starts in one app, continues in another and completes through a payment, download or transfer.
Child exploitation
- NCMEC received 21.3M CyberTipline reports in 2025; online-enticement reports rose 156% to 1.4M, including 80,000+ sextortion reports (NCMEC).
- Financially motivated sextortion averaged 137 reports a day (+37%); NCMEC knows of at least three dozen US teenage boys who died by suicide after sextortion (NCMEC 2025).
- The IWF counted 3,443 AI-generated child sexual abuse videos in 2025, against 13 in 2024 (IWF). About 1 in 17 young people say someone made deepfake nudes of them (Thorn).
Scams and fraud
- The FBI logged 1,008,597 complaints and $20.877B in losses in 2025 (+26%); investment fraud cost $8.6B (HSToday). AI-linked complaints cost $893M (SecureWorld).
- Americans aged 60+ lost $7.75B (+59%) (InvestmentNews). The FTC separately recorded about $15.9B in fraud losses (AARP); the datasets overlap and are never added together.
Pornography, drugs, weapons, harassment and misinformation
- 73% of US teens have watched online pornography; the average age of first exposure is 12, and 58% encountered it without seeking it (Common Sense Media).
- The DEA says cartels target students through social media, and 5 of 10 fentanyl pills it tests contain a potentially lethal dose (DEA).
- Police recovered more than 92,000 ghost guns in 2017–2023 (The Trace); in the UK, 13% of MI5 terrorism subjects are under 18 (ITV).
- Digital piracy — pirated films, music, software and games — is a crime against creators and a common delivery route for malware. (Market data on piracy was not part of this research and is an open item.)
- 46% of US teens have experienced cyberbullying (Pew); 70% of Americans call online misinformation a major threat (Pew).
Workplaces
- The average data breach costs $4.99M globally and about $11.5M in the US (IBM via Infosecurity); the human element is involved in 62% of breaches (Verizon DBIR).
- Insider risk costs $19.5M per organization per year (vendor-sponsored, DTEX/Ponemon); business email compromise cost $3.05B in 2025 (JD Supra).
04
How we got here
No one decided to allow crime. A chain of separate engineering, legal and security choices — each made for stated reasons — left prevention without a natural home: at the edges, after the fact, or nowhere.
| When | Choice | Stated reason | Effect on prevention |
|---|
| 1988 | TCP/IP goals rank survivability first and accountability last; security is absent (Clark 1988) | Military research network of trusted users | No natural place for abuse control in the network core |
| 1981–84 | End-to-end argument: keep the network simple, put functions at the edges (Saltzer, Reed & Clark) | Efficiency and flexibility | Responsibility pushed to endpoints — which, by 2001, its authors conceded could no longer be trusted (Blumenthal & Clark) |
| 1996 | Section 230 immunizes intermediaries and encourages "blocking and filtering technologies that empower parents" (47 U.S.C. § 230) | Innovation, free markets, parental control | Prevention assigned to user-side tools — which stayed optional |
| 1997–2004 | Reno v. ACLU and Ashcroft v. ACLU: filters at "the receiving end" are less restrictive than mandates (Columbia GFoE) | First Amendment | Source-side mandates fail; receiving end preferred |
| 1998–2000 | DMCA § 512(m) and EU E-Commerce Art. 15 bar general monitoring duties (EUR-Lex) | No general monitoring | Notice-and-takedown, after the fact |
| 2002–04 | Microsoft Palladium / NGSCB trusted layer shelved (Tom’s Hardware) | Lock-in and DRM concerns | No trusted OS enforcement layer |
| 2007–18 | OS parental controls arrive late and opt-in: Vista (2007), Screen Time (2018) (Engadget) | User choice | Protection depends on setup and can be bypassed |
| 2021–22 | Apple abandons on-device CSAM scanning (MacRumors) | Bulk-surveillance risk | Detection limited to opt-in Communication Safety |
| 2023–27 | UK Online Safety Act, California AB 1043, UK device-level commitment, Australian codes | Safety by design | The OS becomes a regulated safety layer |
Lessons for PurityOS: earlier OS-level efforts failed over who controls the device and whether it enables surveillance — not over whether safety matters. PurityOS answers both: it is chosen by the people it protects, it analyses personal content on the device, and it publishes what it blocks and why.
05
The solution
PurityOS enforces protection inside the operating system, below every app, browser and message — where it cannot be closed, bypassed or uninstalled by the software it protects against.
Five layers
- Kernel policy engine — system-wide rules for every app, browser, network connection and file.
- Local + cloud AI — local AI analyses personal content on the device in real time; cloud AI tracks new scam domains, dealer marketplaces, disinformation networks and malware and pushes protections to every device, receiving only threat signals such as URLs and file hashes.
- Tamper resistance — verified boot, signed updates and malware containment.
- Private insights — plain-language summaries for the user first, and for guardians or administrators where they have a role.
- Verified identity — every user verified once at setup; children attached to a verified guardian.
Two editions
| PurityOS (installed) | PurityOS Cloud |
|---|
| Runs | On the device (laptops, desktops, tablets) | In PurityOS’s cloud, streamed to any browser or thin client |
| AI | Local AI on personal content; cloud AI on threat signals only | Local and cloud AI in a hosted workspace |
| Privacy posture | Personal content stays on the device — the pattern of Apple’s on-device Communication Safety (Apple) | Content processed in the provider’s cloud; encrypted, isolated per user |
| Best fit | Individuals, families, privacy-first buyers | Schools, employers, shared and older devices |
| Legal note | Vendor likely never gains knowledge of content | Likely a “provider” with CSAM reporting and preservation duties under 18 U.S.C. § 2258A (Cornell LII) |
Profiles
Personal (anyone who chooses a cleaner digital life, with an optional accountability partner), Family (age-based child profiles, guardian alerts), School (central safeguarding policy, on and off campus) and Workplace (fraud, data-loss and malware prevention without keystroke logging or screen recording).
06
Why now: regulation
Regulation is moving safety duties down the stack to the operating system. Mandates that restrict speech remain vulnerable in court; age signals and product choices people make for themselves are proving durable.
- California AB 1043 (operative 1 January 2027) defines an OS provider as any entity that "develops, licenses, or controls" an OS, and requires an age interface at setup and an age-bracket signal to apps; penalties up to $2,500–$7,500 per affected child (Hunton). Because every PurityOS user is verified, PurityOS can supply a higher-assurance signal than the law requires.
- App-store laws: Texas SB 2420 enforceable after the Supreme Court declined to vacate a stay (6 July 2026); Utah from 6 May 2027; Louisiana 1 July 2027; Alabama 1 January 2027 (Lexology; Loeb & Loeb).
- Federal: amended COPPA Rule in force since 22 April 2026 (Federal Register); FTC policy not to pursue age-check-only data collection (FTC); Take It Down Act enforced since 19 May 2026 (FTC); CIPA requires filtering in every E-Rate-funded school (FCC).
- Courts: Free Speech Coalition v. Paxton (2025) upheld age verification for adult sites; about 27 states now have such laws (CRS).
- International: UK device-level nudity protections to be legislated (Biometric Update); Australia’s codes bind OS vendors (eSafety); EU age-verification app targeted for end-2026 (European Commission).
07
Market
All analyst figures are estimates with opaque methods; we show ranges and name sources.
US bottom-up
- Families: about 33.3M US families with children under 18 (Macrotrends / Census). At $144/year the theoretical ceiling is about $4.8B — arithmetic, not demand.
- K-12: 49.3M public students in 2024–25 (K-12 Dive); 88% of schools are 1:1 (IES); bundles at $10–20 per student imply about $0.5–0.9B a year (KyberGate).
- Faith and recovery: Covenant Eyes reports helping 1.7M+ people (Covenant Eyes); the accountability category is roughly $50–100M (inference).
- Employers: Microsoft Purview add-on about $12/user/month (Nightfall); a new entrant likely earns $3–8/seat/month for DLP features alone, so the wedge is managed, locked-down endpoints.
Illustrative obtainable market (US)
| Channel | Assumption | ARR |
|---|
| Families | 0.5% of US families with children (~167K) at $144/yr | ~$24M |
| Schools | 1% of US public K-12 students (~490K) at $10/yr | ~$5M |
| Personal | 50K subscribers at $72/yr | ~$3.6M |
| Total | | ~$30M |
This is a planning scenario driven by named channels, not a forecast.
08
Competition
Incumbents give away improving controls, but each stops at its own ecosystem and documented bypasses persist. PurityOS combines four things no competitor offers together: tamper resistance, cross-device coverage, categories beyond nudity, and verified identity.
| Player | What it is | Where it stops |
|---|
| Apple Screen Time / Communication Safety | Free OS-native controls; nudity blurred by default for under-18s | Apple devices only; Assistive Access bypass reported (Apple Community) |
| Microsoft Family Safety | Free account-level controls | Web filtering only in Edge (Microsoft) |
| Bark, Qustodio, Canopy, Net Nanny | App-layer parental control | Run beside the apps they monitor; can be bypassed |
| Covenant Eyes, Ever Accountable | Accountability apps for faith and recovery users | App / VPN layer; can’t see inside many apps (Canopy) |
| GoGuardian, Securly, Lightspeed, Gaggle | K-12 filtering and monitoring (20–25M students each) | Surveillance backlash and false positives (EFF) |
| Gabb, Pinwheel, Troomi, Bark Phone | Kid-safe phones | Hardware-plus-carrier models — potential licensees |
Comparable transactions
| Company | Event | Value | Source |
|---|
| Aura + Qoria | Merger and ASX listing, July 2026 | >US$300M pro forma ARR; ~32,000 schools | Aura |
| Aura | Series G, March 2025 | $140M at $1.6B (~10.7x ARR) | Aura |
| Qustodio | Acquired, May 2022 | ~$52M; >4M users | Tech.eu |
| Bark | Series C, 2022 | $30M (~$67M total) | PR Newswire |
| GoGuardian | Tiger Global, August 2021 | $200M | Business Wire |
| ObserveIT | Acquired by Proofpoint, 2019 | $225M | SiliconANGLE |
| DTEX | Series E, March 2024 | $50M at >$400M | Axios |
09
Business model and pricing
| Plan | Price | For |
|---|
| Personal | $6/month | Anyone choosing a cleaner digital life; optional accountability partner |
| Family | $12/month (up to 6 people) | Parents and children, age-based profiles |
| Organizations | Custom per seat | Schools and employers; installed or Cloud edition |
Pricing context. Covenant Eyes charges $18/mo or $198/yr; Ever Accountable about $14.99/mo; Bark Premium $14/mo; Canopy $8.33–$12.50/mo (Content Mavericks; Bark; Canopy). Family at $12 is competitive. The evidence suggests Personal at $6 is underpriced for the faith and recovery segment, which already pays $130–200 a year; a paid accountability add-on is one option to test.
Benchmarks. Hard-paywall trials convert at a 10.7% median versus 2.1% for freemium; 17–32-day trials convert at 42.5%; only 9–12% of monthly payers remain after a year (RevenueCat 2026). We favour annual billing, long opt-out trials and direct billing.
Unit economics. Identity verification costs about $1–2 per adult (SpotSaaS) — about 1% of first-year Family revenue. The installed edition’s main running cost is cloud threat intelligence, since local inference runs on customer hardware; PurityOS Cloud adds per-user hosting and inference and is modelled separately.
10
Go-to-market
Distribution matters more than filtering technology. We copy three models that have worked:
- Schools to homes (B2B2C). Bark offers schools its product free and reports 3,400–3,700 districts, then converts parents (Bark). A PurityOS School edition — often PurityOS Cloud on existing Chromebook fleets — converts to Family subscriptions at home.
- Faith and recovery communities. Churches already deploy accountability software congregation-wide (Malay Mail / WaPo). We partner with communities on an opt-in, dignity-first basis.
- OEM and kid-phone partners. Pinwheel earns about $15/month of pure software revenue on partner hardware (TechCrunch). Kid-phone makers and Chromebook OEMs are licensees, not competitors.
- Employers. Lead with fraud, phishing and data-loss prevention on managed fleets in regulated and frontline industries.
- Compliance as a feature. “AB 1043-ready” and CIPA-compliant out of the box.
11
Identity verification
Every PurityOS user is verified once at setup through 1Profile (1profile.com), AW3 Technology’s identity service. Users sign in with Sign in with 1Profile (OAuth 2.0 / OpenID Connect), and 1Profile Verify performs the identity check (document OCR and face match, plus sanctions and PEP screening). Owning the identity layer within AW3 keeps verification costs and data handling under our control. The design follows the evidence on cost, friction and breach risk:
- Wallet credentials first (mobile driver’s licences, EU and Australian digital ID), falling back to a 1Profile Verify document check with zero retention of ID images on PurityOS devices. Google Wallet already supports zero-knowledge age proofs (Google).
- Store attestations, never documents: verified human, age band, role and a pseudonymous ID, bound to the device — as LinkedIn’s Persona integration shares only the result (CleanUpTheInternet).
- Verify the guardian, not the child. COPPA requires verifiable parental consent, not child ID; the FTC declined facial age estimation as a consent method in 2024 (Baker Hostetler).
- Keep raw IDs out of support. The 2025 Discord leak of ~70,000 IDs came through a support vendor’s appeals path (Techdirt).
What we claim. Verification raises the cost of disposable-account abuse, enables accountability through legal process, sets age-appropriate defaults and stops adults posing as children. We do not claim a measured percentage reduction in crime: South Korea’s mandatory real-name law did not demonstrably reduce abuse and was struck down (Korea Herald). Verification inside a product people choose is closer to the voluntary model that research found effective (Cho & Kwon 2015).
12
Product and technology roadmap
We build on proven open-source foundations rather than an OS from scratch — Firefox OS and Windows Phone failed over the app gap (Engadget). The technology strategy (docs/research/reports/PurityOS OS foundation strategy.md) recommends three stages: a protection layer on existing devices first, then a hardened AOSP fork as the first OS PurityOS owns, then a signed Linux image that serves both the desktop edition and PurityOS Cloud. ChromiumOS is ruled out: Google’s successor, Googlebook OS, shipped closed source in October 2026 and ChromeOS updates end in 2034.
| Phase | Milestones (planning assumptions) |
|---|
| 1 — PurityOS Shield (months 0–12) | Protection layer on existing devices: Android device-owner app with locked VPN/DNS and on-device AI; Chromebook management and DNS filtering for schools; Windows and macOS agents; limited iOS app. PurityOS Cloud begins as a filtered cloud browser. Builds the shared policy engine, AI models, threat intelligence and 1Profile sign-in and verification. |
| 2 — PurityOS for phones (months ~9–30) | Hardened AOSP fork without Google services, in the style of GrapheneOS, with a curated app store; launches on Pixel 8 and later (the only commodity phones that allow relocking the bootloader with our own keys), adding Motorola once its compatible devices ship in 2027. Independent classifier accuracy audit; AB 1043 age signal. |
| 3 — PurityOS for laptops and Cloud (months ~24–48) | Signed, image-based Linux desktop (Fedora / CentOS Stream bootc) for Intel and AMD laptops first; the same image streamed to a browser becomes the full PurityOS Cloud desktop. Firmware partnerships with laptop makers for hard tamper resistance. |
| 4 — Scale | Workplace edition (fraud, DLP, malware); OEM and kid-phone licensing; UK, Australia and EU compliance; published transparency reports; optional seL4-based policy core. |
Engineering cost, hardware requirements for local AI and Cloud hosting costs are open items to be estimated internally.
13
Risks and mitigations
| Risk | Evidence | Mitigation |
|---|
| On-device scanning backlash | Apple abandoned CSAM scanning in 2022; “Bugs in our Pockets” (arXiv) | Detect and block locally; no vendor or police reporting by default on installed devices; audited category lists and transparency log |
| Classifier false positives | Google flagged a father over medical photos (9to5Google) | Warn or blur before block; appeals; allowlists for medical, educational and health content; published error rates |
| School surveillance backlash | CDT: 29% of LGBTQ+ students report outing (CDT) | No identity keyword lists; no after-hours police routing; alert content stays on device |
| Misinformation category | Most politically contested category | Label rather than block by default; transparent third-party sources; configurable |
| CSAM reporting status | Duty attaches on “actual knowledge” (18 U.S.C. § 2258A) | Edition-specific legal opinions; Cloud edition built for reporting and preservation duties; NCMEC partnership |
| Identity data breach | Tea app and Discord breaches (The Record) | Attestations only; vendor zero-retention; written biometric consent |
| Employer monitoring liability | CNIL fined Amazon €32M (CNIL) | Block, don’t log; no keystroke or screen capture; DPIA and notice templates |
| OS execution | Tamper resistance needs locked bootloaders and attestation | Staged plan: protection layer first, then AOSP on Pixels with relocked bootloaders, then Linux; be explicit that tamper resistance is prevented on relocked phones and only detected on ordinary PCs; Android Enterprise and ChromeOS management paths |
14
Financial plan
The model is built from named channels, not a share of TAM. All figures below are planning assumptions to be validated in pilots.
| Year 1 | Year 2 | Year 3 | Year 4 | Year 5 |
|---|
| Stage | Build & pilots | Launch | Channels | Scale | Scenario target |
| Family subscribers | Pilot | 15K | 50K | 110K | 167K |
| Personal subscribers | Pilot | 5K | 15K | 30K | 50K |
| Students (school seats) | Pilot | 50K | 150K | 300K | 490K |
| Illustrative ARR | — | ~$3M | ~$9M | ~$20M | ~$30M |
At the Aura benchmark of roughly 10.7x ARR (ImpactAlpha, secondary), a $30M ARR business would be valued around $300M — treated as a realistic, not peak, multiple given Aura’s subsequent down round and weak listing.
15
Team
PurityOS is developed by AW3 Technology Venture Studio, which builds and launches technology companies from the ground up.
Key hires: OS / platform security lead, on-device ML lead, trust & safety and policy lead, education partnerships lead.
16
Funding and use of funds
PurityOS is raising a seed round. Amount and terms to be confirmed by AW3 Technology Venture Studio.
| Use of funds | Share |
|---|
| OS platform, security and tamper resistance | 40% |
| Local and cloud AI, independent accuracy audits | 25% |
| Identity, privacy and compliance (AB 1043, COPPA, legal opinions) | 10% |
| Pilots and partnerships (schools, faith communities, OEMs) | 20% |
| Operations | 5% |
17
Open questions
- Verify key figures against primary sources (AB 1043 statutory text, canonical IC3 report, IBM, DBIR, DTEX).
- Commission independent precision and recall testing for each classifier category.
- Estimate hardware requirements for local AI and hosting costs for PurityOS Cloud.
- Obtain legal opinions on § 2258A provider status for each edition, and FERPA / COPPA consent for school deployments.
- Validate pricing, conversion and churn in pilots.
- Research digital piracy losses and enforcement (not covered in the current evidence base).
Full evidence base with every source: docs/research/reports/PurityOS business plan research.md.