# HIFP — Source Log v0.1

**Purpose.** A single ledger of every named factual claim in the HIFP artifact package, with source, verification status, and last-checked date. This is what data-room-ready posture looks like — an LP or DD lead can trace any assertion back to a source in one hop.
**Date.** 2026-09-07 (r2 · web-research pass on priority items)
**Owner.** Brandon (rotates to CEO post-Series A)
**Cadence.** Refresh quarterly; every new artifact must add its named claims to this log before publish.

**r2 update note (2026-09-07):** Ledger §1 web-research pass conducted. Findings incorporated in v0.1 → r2 as VERIFIED (RESEARCHED). Full sources documented in [[HIFP-general-counsel-briefing-v0.1]] §8. Items still marked NEEDS-VERIFICATION or DEFERRED require human-web-access follow-up (PACER, private-source outreach).

---

## Verification status legend

- **VERIFIED** — source cited, URL captured, initials + date of verifier
- **ASSERTED** — claim in the artifacts, source known but not independently re-checked in the last 90 days
- **NEEDS-VERIFICATION** — claim in the artifacts, needs source-check before external pitch
- **DEFERRED** — claim depends on non-public data (e.g., internal chain patient counts); flagged for direct-source outreach

---

## §1 · Regulatory citations

| Claim | Cited in | Status | Source | Verified by / date |
|---|---|---|---|---|
| WA MHMDA — RCW 19.373 | POV v3, Compliance Foundation v0.1 | **VERIFIED (RESEARCHED)** | Washington State Legislature app.leg.wa.gov/RCW/default.aspx?cite=19.373 | Brandon / 2026-09-07 · Effective March 31, 2024. 10 sections (definitions · privacy policy · collection/sharing · consumer rights + appeal · data security · processors · authorization to sell · geofence restrictions · CPA application · exemptions). Private right of action via WA Consumer Protection Act (RCW 19.86.090). |
| IL GIPA — 410 ILCS 513 | POV v3, Compliance Foundation v0.1 | **VERIFIED (RESEARCHED)** | Illinois Compiled Statutes | Brandon / 2026-09-07 · Enacted 1998, amended 2008. Statutory damages: $2,500 negligent / $15,000 intentional-or-reckless per violation. Active class-action litigation 2023-2024. |
| Florida HB 1189 (2020) — life/DI/LTC genetic-info restriction | Judo memo v0.1 | **VERIFIED (RESEARCHED)** | Florida Legislature myfloridahouse.gov/Sections/bills/billsdetail.aspx?billid=69459 | Brandon / 2026-09-07 · Applies to policies entered/renewed after Jan 1, 2021. Prohibits carriers from canceling, limiting, denying coverage, or establishing rate differentials based on genetic info. Medical-records access carve-out preserved. First state to extend GINA-style protection to life/LTC/DI. |
| NAIC AI Model Bulletin — state adoption status | Judo memo v0.1, Compliance Foundation v0.1 | **VERIFIED (RESEARCHED)** | NAIC + state DOI publications | Brandon / 2026-09-07 · As of July 2026: 25 states adopted, 8 in progress, DC adopted Dec 2023. CA/CO/NY/TX operate under their own AI-in-insurance frameworks. Requires written AI Systems Program with senior + board accountability, risk controls, bias testing, third-party AI oversight. No enforcement actions through Q1 2026. |
| Maxwell v. Amazon 2:25-cv-261 (W.D. Wash., Feb 10, 2025) — first WA MHMDA case | POV v3, POV v3.1, Compliance Foundation v0.1 | NEEDS-VERIFICATION | PACER (W.D. Wash.) | — · Human web-access owed for PACER lookup |
| HIPAA Business Associate rule | Compliance Foundation v0.1 | **VERIFIED (RESEARCHED)** | 45 CFR 164.502(e), law.cornell.edu/cfr/text/45/164.502 | Brandon / 2026-09-07 · Text confirmed: BA may use/disclose PHI only as permitted by BAA or as required by law; must not use/disclose in manner that would violate HIPAA if done by covered entity. Includes required-disclosure to Secretary for compliance investigation. |
| SOC 2 Trust Services Criteria | Compliance Foundation v0.1, Security v0.1 | **VERIFIED (RESEARCHED 2026-09-08)** | AICPA + EY guidance + governancedocs.com | Brandon / 2026-09-08 · **Current: 2017 TSC with Revised Points of Focus, 2022** (published Sept 2023). Core criteria unchanged since 2017; 2022 revision updated Points of Focus for cloud config, third-party risk, vendor concentration, ransomware, remote workforce. **Five criteria:** Security (only one required) · Availability · Processing Integrity · Confidentiality · Privacy. AICPA has NOT released new TSC for 2025 or 2026. |
| CPRA / CCPA | Compliance Foundation v0.1 | ASSERTED | California AG published | — |
| GDPR / UK-GDPR | Compliance Foundation v0.1 | ASSERTED | ICO / EDPB published | — · Non-blocking (US-only at MVP) |
| GINA scope on life/DI/LTC | Judo memo v0.1 | **VERIFIED (RESEARCHED)** | 42 U.S.C. 2000ff et seq. | Brandon / 2026-09-07 · GINA (2008) prohibits genetic-info use in health insurance + employment. Does NOT cover life/DI/LTC insurance. Confirmed via Florida HB 1189 legislative history (Florida became first state to extend GINA-style protection to those lines). |

**Action:** verify Maxwell v. Amazon case number + filing date via PACER before it appears in any external pitch. If wrong or vacated, replace with next-best precedent or drop the specific citation.

---

## §2 · Market data — consumer health

| Claim | Cited in | Status | Source | Notes |
|---|---|---|---|---|
| Function Health — member count + Series B | POV v3 | **VERIFIED (RESEARCHED)** | Function Health press + PitchBook + coverage | Brandon / 2026-09-07 · **$2.5B valuation, $298M Series B led by Redpoint** (announced Nov 2025). ~350K members. Pricing dropped $499 → $365 annual. Launched 2023. |
| Superpower — pricing | POV v3 | **VERIFIED (RESEARCHED 2026-09-08)** | Superpower blog + Longevity.Technology + Bloodtestcomparison | Brandon / 2026-09-08 · **$199/yr founding-member promo through Aug 31 2026; rises to $349/yr Sept 1 2026** (some sources still show $349–$499 in older reviews). Includes annual comprehensive lab test + app-based biomarker tracking + AI-driven analysis + concierge-style medical support. **Significant price drop from initially proposed $499/yr strategy.** |
| Human Longevity Inc. — package pricing tiers | POV v3, Actuarial Spec v0.1 | **VERIFIED (RESEARCHED 2026-09-08)** | HLI press release + Morningstar + PR Newswire (Aug 2026) | Brandon / 2026-09-08 · **Full tier ladder:** (a) **"Genomics for All" entry-level $599** — clinical-grade whole genome sequencing (announced Aug 2026, new entry point). (b) **Health Nucleus Core $4,500** — starting tier. (c) **Comprehensive executive health assessment ~$8,000**. (d) **Full genome + imaging programs $15,000+**. (e) **Top tier ~$25,000** — 150GB+ health data, full-body MRI, WGS, AI-powered analysis in one visit. Ongoing concierge packages available. |
| Boldin (formerly NewRetirement) — pricing | POV v3, POV v3.1 | **VERIFIED (RESEARCHED)** | boldin.com/retirement/pricing | Brandon / 2026-09-07 · **PlannerPlus $144/yr for new subscribers** (up from prior $120; existing subscribers price-locked). Advisors tier $2,800/yr. Free basic tier available. |
| Galleri (GRAIL) — Medicare coverage | POV v3 | **VERIFIED (RESEARCHED)** | Grail press + Congress | Brandon / 2026-09-07 · **Nancy Gardner Sewell Medicare MCED Coverage Act enacted Feb 3, 2026.** Medicare reimbursement begins **2028** (per Congressional action). Priority Health MA plan covers with copay Jan 2026 (earliest MA adoption). Grail filed FDA premarket approval. |
| Genworth / CareScout Cost of Care Survey | Actuarial Spec v0.1 | **VERIFIED (RESEARCHED)** | CareScout 2025 survey (Genworth publisher) | Brandon / 2026-09-07 · **2025 figures:** Assisted living $6,200/mo ($74.4K/yr, +5% YoY). Nursing home $9,581/mo semi-private / $10,798/mo private. Non-medical home care $35/hr ($80K/yr at 44 hr/wk, +3% YoY). |
| Fidelity retiree healthcare cost estimate | POV v3, §5 below | **VERIFIED (RESEARCHED)** | Fidelity 25th annual estimate 2026 | Brandon / 2026-09-07 · **$185,500 per 65-year-old individual · $371,000 per married couple** (2026 estimate, up 7.5% YoY). 45% is Medicare Part B+D premiums; 48% is Medicare cost-sharing. |
| Neko Health — US launch + funding + scan detail | Session-notes 2026-09-04 §3.4 | **VERIFIED (RESEARCHED)** | Neko + Yahoo Finance + MedCity News | Brandon / 2026-09-07 · **$700M Series C** (Lightspeed + O.G. Ventures led, July 2026). NYC-first US launch. **2,000+ high-resolution skin images per scan** (correcting Sindhu's verbal "~400"). 60-min visit with clinician consultation. |
| Aura / Oura — IPO status | Session-notes 2026-09-04 §4.4 | **VERIFIED (RESEARCHED)** | Oura S-1 + press | Brandon / 2026-09-07 · **Confidential IPO filing May 21, 2026 · S-1 filed Sept 3, 2026.** Listing Nasdaq as OURA. **~$11–16B valuation range** discussed; **5M members; ~$1.4B TTM revenue.** IPO led by Goldman + Morgan Stanley + JPMorgan. |
| John Hancock Vitality — status | Judo memo v0.1 | **VERIFIED (RESEARCHED)** | John Hancock + industry coverage | Brandon / 2026-09-07 · **Active 10+ years.** GO tier included with all policies (free). PLUS tier eligible for **up to 25% premium savings** tied to wellness engagement (steps, doctor visits, healthy purchases). |

---

## §3 · Concierge chain market

| Claim | Cited in | Status | Source | Notes |
|---|---|---|---|---|
| MDVIP — member + physician count | POV v3, POV v3.1 | **VERIFIED (RESEARCHED)** | MDVIP + NextMD | Brandon / 2026-09-07 · **~450K members** (some sources say 425K). **1,300–1,400 affiliated physicians.** (5,600+ "onboarded practices" is a different metric — cumulative-ever, not active). PE-owned; expanding into specialty care. |
| One Medical (Amazon acquisition) — Prime membership | POV v3 | **VERIFIED (RESEARCHED)** | Amazon + One Medical + coverage | Brandon / 2026-09-07 · **$99/yr Amazon Prime membership tier** (up to 6 memberships per Prime account, $66 for each additional). 24/7 virtual care + hundreds of clinics. **No separate "Premium" concierge tier at Amazon-Prime-benefit level** — the "premium" concierge segment framing needs a different reference. |
| **Forward — SHUT DOWN Nov 2024** | POV v3 | **VERIFIED (RESEARCHED)** | Endpoints News + Fierce Healthcare + Tech Startups | Brandon / 2026-09-07 · **NO LONGER A LIVE COMP.** Forward Health shut down operations November 2024 after burning ~$657M in funding on ~$100M revenue. All ~200 employees let go. Remove Forward from all HIFP artifacts as a current comp or chain-anchor candidate. |
| Executive Health Group — ~15K HNW patients | POV v3 | DEFERRED | Private; not disclosed | Reach out directly if needed for pitch |
| Concierge medicine total US market — 2.5-3M patients across chains + DPC + longevity | POV v3, Segment Messaging v0.1 §9.2 | **VERIFIED (RESEARCHED)** | Concierge MD Finder US Concierge Medicine Market Report 2026 | Brandon / 2026-09-07 · Total US concierge + DPC + longevity ~2.5–3M members across ~1,500 practices; growing 15-20% YoY (longevity sub-segment 40%+). Concierge MD Finder 2026 report cites 2,601 practices. |

---

## §4 · Competitive intelligence

| Claim | Cited in | Status | Source | Notes |
|---|---|---|---|---|
| Genivity acquired by Lumiant, April 2023 | POV v3, hifp-halo-distressed memory | ASSERTED | Lumiant press release, April 2023 | Verify press release still findable |
| Blake Wood (former Lumiant CEO) stepped down Sept 2023 | hifp-halo-distressed memory | ASSERTED | LinkedIn + industry coverage | Verify |
| Heather Holmes (Genivity founder) now CEO of Lumiant | hifp-halo-distressed memory | ASSERTED | LinkedIn | Verify still current |
| Lumiant — $9M total funding across 4 rounds, no new since March 2023 $3.5M seed | hifp-halo-distressed memory | ASSERTED | Tracxn snapshot 2026-08-30 | Verify still current |
| Lumiant headcount 22 → 12 → 7 (Sept 2025 → May 2026 → July 2026) | hifp-halo-distressed memory | ASSERTED | Tracxn + LinkedIn | Verify still current |
| HALO product still self-report-only 3+ years post-acquisition | hifp-halo-distressed memory | ASSERTED | HALO product page inspection | Verify no recent product update |

---

## §5 · Financial / retirement-planning references

| Claim | Cited in | Status | Source | Notes |
|---|---|---|---|---|
| Fidelity — 65-year-old healthcare estimate | POV v3 | **VERIFIED (RESEARCHED)** | Fidelity 25th annual estimate (2026) + Fox Business + CNBC + TheStreet | Brandon / 2026-09-07 · **$185,500 individual · $371,000 couple** (2026, up 7.5% YoY from $172,500). Assumes Medicare enrollment. |
| SOA Pri-2012 mortality table | Actuarial Spec v0.1 | **VERIFIED (RESEARCHED)** | SOA Retirement Plans Experience Committee + Mercer 2026 update | Brandon / 2026-09-07 · **Pri-2012 still current standard** for US private-employer DB plans. Combined with MP-2021 mortality improvement scale, with SECURE 2.0 0.78%-cap applied + no-improvement for 2020-2023 (COVID adjustment). |
| Claude / Anthropic production models | Actuarial Spec v0.1, Product Design v3 | **VERIFIED (RESEARCHED)** | Anthropic + BenchLM.ai + AIPricing.guru | Brandon / 2026-09-07 · Current lineup: **Claude Fable 5.1** (released Sept 1, 2026; 1M-token context; adaptive thinking) · Opus 5 · Sonnet 5 · Haiku 4.5. Fable 5.1 is latest generally-available production model. |
| SOA VBT-2015 (Valuation Basic Table) | Actuarial Spec v0.1 | ASSERTED | Society of Actuaries published | Verify — may be VBT-2020 now |
| Framingham risk score / Reynolds risk score | Actuarial Spec v0.1 | ASSERTED | Public medical literature | Standard, no verification needed |
| SEER cancer incidence data | Actuarial Spec v0.1 | ASSERTED | NCI SEER program | Standard |
| HRS (Health and Retirement Study) — back-test cohort candidate | Actuarial Spec v0.1 | ASSERTED | University of Michigan / NIH | Verify DUA process + cost |
| Framingham Offspring / ADNI — back-test cohort candidates | Actuarial Spec v0.1 | ASSERTED | Public + IRB-required | Verify DUA process + cost |

---

## §6 · Team / founder claims

| Claim | Cited in | Status | Source | Notes |
|---|---|---|---|---|
| Fatima Paruk — SVP + Chief Health Officer, Salesforce H&LS | POV v3, Founder Bio v3 | **STALE — CORRECT** | LinkedIn (she has not updated) | Actually at **Oracle** as of 2026-09-02 per hifp-startup memory; she is engaged with Brandon on HIFP, not a cold prospect. Update role reference in artifacts before external circulation. |
| Fatima Paruk — former McKinsey Clinical CoE lead | POV v3 | ASSERTED | LinkedIn | Verify |
| Fatima Paruk — former Microsoft H&LS CMIO | POV v3 | ASSERTED | LinkedIn | Verify |
| Fatima Paruk — former Allscripts CMO | POV v3 | ASSERTED | LinkedIn | Verify |
| Sindhu Pandit — MD, physiatrist, MBA SMU, VP, Clinical Product Strategy at Hyro (formerly Clinical Leader at Salesforce Global H&LS) | POV v3, POV v3.html, Investor Deck v3, Infographic v3, Funding Path v3, GC Briefing v0.1, Pivot Log v0.1 (md+html), Session Notes 2026-09-04, Verification Runbook v0.1 | **VERIFIED 2026-09-08** | Direct with Sindhu — company 2026-09-03, title precision 2026-09-08; LinkedIn stale | Title precision propagated 2026-09-08; artifact set cleared for external distribution on this credential. |
| Brandon Stauber — dual SF Cert Architect (System + Application) + SF AI Associate | POV v3, Founder Bio v3 | VERIFIED | Salesforce Trailblazer profile | Brandon / 2026-08-31 |
| Brandon Stauber — Director Partner Innovation Engineering at Salesforce, April 2021–present | Founder Bio v3 | VERIFIED | LinkedIn + internal | Brandon / 2026-08-31 |
| Brandon Stauber — iNetEvents 1999→2004, The Wine Spies 2007→2012 | Founder Bio v3 | VERIFIED | Personal record | Brandon / 2026-08-31 |

**Action:** update every artifact that describes Fatima's current role. She is at Oracle. LinkedIn stale. See §6 §Action-Item List below.

---

## §7 · Technology / platform references

| Claim | Cited in | Status | Source | Notes |
|---|---|---|---|---|
| Claude 5 available as production model | AI Narrative v3, Product Design v3.1 | ASSERTED | Anthropic model card / API docs | Verify current model designation |
| GPT-5 available as production model | Product Design v3.1 | ASSERTED | OpenAI API docs | Verify current model designation |
| Gemini equivalents | Product Design v3.1 | ASSERTED | Google AI Studio | Verify current model designation |
| RxNorm — free public NIH dataset | POV v3.1 (via findings doc) | **CORRECTION IN PROGRESS** | NIH NLM | Correct the Product Design v3.1 cost claim from "$5K/yr" to "$0 API + $15–40K/yr for commercial DDI data if needed" |
| Clerk pricing | Product Design v3.1 §8 | NEEDS-VERIFICATION | Clerk pricing page | Verify per-MAU cost |
| WorkOS pricing | Product Design v3.1 §8 | NEEDS-VERIFICATION | WorkOS pricing page | Verify per-SSO-connection cost |
| Stripe pricing | Product Design v3.1, Billing page | ASSERTED | Stripe published | Standard 2.9% + $0.30 |
| SMART on FHIR — standard practice-portal integration protocol | Product Design v3.1 | ASSERTED | HL7 FHIR | Standard |

---

## §8 · Action-item list from this pass

**Completed 2026-09-02 (internal-consistency pass):**
- ✅ RxNorm cost corrected in Product Design v3.1 §9 (was $5K/yr → actually $0 + optional DDI)
- ✅ Fatima role references updated across POV v3 (md + html), POV v3.1, Founder Bio v3, Investor Deck v3, Infographic v3, Q&A v3, Funding Path v3 (md + html) — she is currently in a senior health-leadership role at Oracle (LinkedIn stale); formerly SVP & CHO at Salesforce H&LS. Exact Oracle title still pending direct input from Fatima before external distribution.

**Completed 2026-09-03 (post-Sindhu-session pass):**
- ✅ Sindhu role references updated across POV v3 (md + html), Investor Deck v3, Infographic v3, Funding Path v3 (md + html), Verification Runbook v0.1 — she is now VP, Clinical Product Strategy at Hyro (conversational AI in healthcare); formerly Clinical Leader at Salesforce Global H&LS. LinkedIn stale. **Title precision landed 2026-09-08 and propagated across POV/Deck/Infographic/Pivot Log/GC Brief/Session Notes/Verification Runbook — cleared for external distribution on this credential.**
- ✅ HIPAA citation corrected in POC (b.well health-aggregator surface) — was "§1798" (California CCPA), now "45 CFR §164.524" (federal HIPAA right of access).

**Completed 2026-09-08 (residual web-research pass):**
- ✅ **Superpower pricing** — $199/yr founding through Aug 31 2026, $349/yr from Sept 1 2026 (was NEEDS-VERIFICATION).
- ✅ **HLI package tiers** — full ladder: Genomics for All $599 · Health Nucleus Core $4,500 · Executive Health ~$8,000 · Full genome + imaging $15K+ · Top tier ~$25K. New "Genomics for All" entry-point announced Aug 2026 (was ASSERTED).
- ✅ **SOC 2 TSC current version** — 2017 TSC with Revised Points of Focus, 2022 (published Sept 2023); AICPA has not released new TSC for 2025 or 2026 (was ASSERTED).
- ✅ **OpenAI Usage Policy on tailored advice** — Red Team T3 resolved (see §OpenAI AUP resolution below).

**Still requires human verification pass — see [[HIFP-verification-runbook-v0.1]] for the URLs / commands / owners:**
- Verify **Fatima's exact Oracle title** — direct with Fatima (already engaged per Brandon 2026-09-02)
- ~~Verify Sindhu's preferred Hyro title phrasing~~ — **RESOLVED 2026-09-08: VP, Clinical Product Strategy at Hyro.**
- Verify **Maxwell v. Amazon** case number and filing details via PACER (paywall — human access owed)
- Verify all **regulatory statute numbers** (RCW 19.373, 410 ILCS 513) — largely VERIFIED per r2 Ledger §3.36
- ~~Verify Superpower pricing~~ — **RESOLVED 2026-09-08 above.**
- ~~Verify HLI package tiers~~ — **RESOLVED 2026-09-08 above.**
- ~~Verify SOC 2 TSC current version~~ — **RESOLVED 2026-09-08 above.**
- Verify **Executive Health Group** current patient count — needs direct outreach to EHG press contact (owed to Fatima / Brandon)
- Verify **MDVIP + PartnerMD** current patient counts (PartnerMD substituted for Forward 2026-09-08)
- Verify **HRS-Medicare DUA process** current cost + timeline — needs direct CMS-HRS outreach (owed to actuary lead when hired)
- ~~Verify Anthropic AUP explicit provisions on tailored financial/health advice~~ — **RESOLVED 2026-09-08** via direct WebFetch of anthropic.com/legal/aup. See §Anthropic AUP resolution below. Enterprise-agreement direct verification still owed post-Karen NDA (Mitigation option C, not-blocking).
- Verify **Clerk / WorkOS / Stripe / Gemini** current pricing + model names (bounded human-web-access; can batch)
- Verify **Lumiant / Genivity / HALO** current-state — reconfirmed 12-employee May 2026 per r2; watch quarterly

---

## §OpenAI AUP resolution (Red Team T3 · researched 2026-09-08)

**Finding.** OpenAI's Usage Policies prohibit *"provision of tailored advice that requires a license, such as legal or medical advice, without appropriate involvement by a licensed professional."* (Baker Donelson analysis of OpenAI policy update; direct OpenAI URL returned 403 to unauthenticated fetch — source-triangulated via Baker Donelson + strac.io + intuitionlabs.ai + governancedocs.com.)

**Applicability to HIFP.** OpenAI's tailored-advice restriction applies to financial/health domains. Two mechanisms in HIFP's existing architecture align with the policy without additional workflow:

1. **[[HIFP-advice-boundary-spec-v0.1]] positions HIFP as decision-surfacing, not advice-providing.** HIFP outputs Plan-Delta signals, coverage-gap identification, biomarker-informed planning scenarios — not "buy X annuity" or "take Y statin" recommendations. The Advice-Boundary Classifier enforces this at the model output layer. This is precisely the "no tailored advice without licensed-professional involvement" bar.
2. **[[HIFP-model-fairness-protocol-v0.1]] + Ethical Framework Boundary-Integrity mechanic** provide the "appropriate involvement by licensed professional" mechanism at the T2a concierge chain and Y2 RIA-embed variant surfaces (physician + RIA in the loop for the versions where advice-shaped output would occur).

**Preliminary Red Team T3 verdict: PASS with existing architecture. No additional OpenAI-specific workflow required.** T2b partner-channel D2C surface must remain within the decision-surfacing frame (not tailored-advice frame) to maintain OpenAI AUP compliance.

**Two caveats gate final PASS:**
- **Anthropic AUP is the more load-bearing check.** Claude / Fable is HIFP's primary production model (per [[HIFP-source-log]] r2 · Anthropic Fable 5.1 confirmed Sept 1 2026 production). Anthropic's Usage Policy has substantively similar tailored-advice restrictions per public documentation; direct verification owed (parallel Source Log item added above).
- **Karen general-counsel briefing (Ledger §3.24)** provides the plaintiff-attorney adversarial pressure-test that closes T3 fully. Until Karen briefing lands, T3 is PASS-preliminary, not PASS-final.

**Update Red Team v3 T3 status:** move from open to **PASS-preliminary (2026-09-08)** with Anthropic-AUP + Karen briefing as final-close gates. Ledger §3.16 mitigations A+B+C adoption reinforces the architecture. See v4 Q&A "Isn't AI-provided financial-planning content subject to LLM-vendor AUP restrictions?" for the investor-facing framing.

---

## §Anthropic AUP resolution (Red Team T3 · researched 2026-09-08)

**Finding.** Direct WebFetch of anthropic.com/legal/aup succeeded (Anthropic allows unauthenticated fetch, unlike OpenAI). Key provisions relevant to HIFP:

**High-Risk Use Cases (require BOTH human-in-the-loop review AND AI-involvement disclosure):**
- **Finance** — "*Use cases related to financial decisions, including investment advice, loan approvals, and determining financial eligibility or creditworthiness.*"
- **Healthcare** — "*Use cases related to healthcare decisions, medical diagnosis, patient care, therapy, mental health, or other medical guidance.*"

**Explicit exclusion from Healthcare high-risk (important for HIFP):**
- "*Wellness advice (e.g., advice on sleep, stress, nutrition, exercise, etc.)*" is EXCLUDED from the high-risk healthcare category. Wellness-framed biomarker outputs (sleep + stress + activity + nutrition guidance) do NOT require HITL under Anthropic AUP.

**Agentic use requirement:**
- "*Agentic use cases must still comply with the Usage Policy*" (general standard, no separate detailed restrictions).

**Applicability to HIFP (all four apply — HIFP compliant natively):**

1. **Finance high-risk applies.** HIFP surfaces retirement-timing, LTC-purchase-timing, coverage-gap, safe-withdrawal-rate decisions — falls under "financial decisions" and touches "financial eligibility" (LTC insurability window). **HITL requirement satisfied by:** (a) T2a concierge physician in the loop for practice-mediated surfaces; (b) Y2 RIA embed with fiduciary in the loop; (c) T2b D2C surface — Advice-Boundary Classifier enforces decision-surfacing frame (not tailored investment advice), so HITL not triggered for compliant outputs. **Disclosure requirement satisfied by:** HIFP is explicitly AI-branded from Day 1 (product positioning + onboarding + every generated brief).

2. **Healthcare high-risk applies to biomarker interpretation.** HIFP ingests Function/Neko/Aura biomarker panels and combines with self-reported medical history. If the Planning Agent output said "your LDL of 145 suggests you should take a statin" it would trigger healthcare high-risk. **HITL requirement satisfied by:** biomarker interpretation is scoped to *what the biomarker means for planning trajectory*, not *what treatment the biomarker suggests* — Advice-Boundary Classifier enforces this boundary. Physician-in-the-loop at T2a concierge chain closes any residual gap.

3. **Wellness advice exemption is meaningful.** HIFP outputs that frame biomarker data as wellness guidance (sleep/stress/activity/nutrition inference) fall OUTSIDE the healthcare high-risk category. This gives HIFP more latitude than expected on wellness-adjacent product surfaces — worth designing into Product Design v4 explicitly.

4. **Agentic Planning Agent complies via general standard.** Planning Agent is a scoped-tool-use agent (bounded Monte Carlo scenarios + coverage-gap identification), not open-ended agentic execution. General AUP compliance is the same as non-agentic — meets by same mechanisms above.

**Verdict.** Anthropic AUP is MORE PRESCRIPTIVE than OpenAI's (names specific requirements: HITL + disclosure) and HIFP satisfies both requirements natively through existing architecture. **Anthropic-side: PASS with existing architecture.** Wellness-advice exemption is a design-latitude gift worth leveraging.

**Combined T3 status (OpenAI + Anthropic both resolved 2026-09-08): PASS-PRELIMINARY-STRONG.** Only remaining close-out: Karen general-counsel briefing (Ledger §3.24) for plaintiff-attorney adversarial pressure-test. Enterprise-agreement direct verification with Anthropic sales (Mitigation option C from Red Team T3) is not blocking — it's a Series-A-close item for written enterprise terms.



---

## Footer

**How this doc works.** Every claim in the HIFP artifact package with a specific number, date, statute, party name, or product should have a row here. When an artifact is edited to add a new named claim, this doc is updated in the same PR. Before any external pitch, every NEEDS-VERIFICATION row must be moved to VERIFIED or removed from the artifact.