AI Dispatch — The Assurance Gap, October 2, 2026

# AI Dispatch — The Assurance Gap, October 2, 2026

**Dek:** Insurance and indemnity: 122 of 123 human-scored published entities (99.2%) still sit at zero or one on `insurance_indemnity` — 119 of them at a flat zero. This week's assurance development is also about risk transfer, and it is the same shape: Richpanel's data processing agreement now commits the vendor to indemnify customers and report breaches within five business days — real terms, reviewed and cited below — but its public materials still name no insurer, no carrier, and no coverage limit for anything the agent itself does.

---

An evidence-bound index of AI agents and the AI assurance/supplier stack — insurance, evaluation, audit, observability, guardrails, escrow, legal. Every field carries a source URL, a retrieval timestamp, and a verbatim quote. No estimates, no growth-hack framing.

## The assurance gap, in real numbers

Queried directly against the live database, 2026-10-02, human-scored dimensions only (excluding the automated `spark-verifier` pass, which cannot write a 0 or a 3 and would overstate coverage if blended in). Read from `scorecards_resolved`, not `scorecards` — the latest score per entity per dimension, published entities only:

- **insurance_indemnity**: 123 scored, **122 (99.2%) sit at 0-1** — 119 at a flat 0, 3 at 1. Exactly one published entity scores above 1: Armilla AI, an AI insurance vendor, at a 2 — the rubric's top score requires a named carrier, a published coverage limit and a customer-indemnity term, and Armilla is not scored as carrying either a limit or an indemnity term. - **outcome_settlement**: 118 scored, 114 (96.6%) sit at 0-1; 108 at a flat 0. - **audit_trail**: 123 scored, 106 (86.2%) sit at 0-1; 74 at a flat 0.

`compliance` and `runtime_governance` — not part of the standing three, but worth naming as contrast — sit at 0-1 for 86.9% of 145 scored entities and 72.1% of 122, respectively. Buyers can more often find a certification claim or an approval step than they can find risk transfer.

Full five-dimension human-scored coverage: **118 of 802 published entities (14.7%)**. More than six of every seven published entities still carry an incomplete scorecard.

## This week's assurance/insurance development: a DPA grows an indemnity clause, not an insurance policy

Richpanel, an e-commerce customer-service agent, added a vendor-indemnity clause and a breach-notification deadline to its data processing agreement, according to a copy this outlet reviewed and published as a wire story on 2026-09-28. The DPA states the vendor "shall indemnify and hold harmless Company and its affiliates, parents, subsidiaries, employees, officers, contractors, and agents from and against any third-party claims arising from (a) a breach of Service Provider's representations or warranties under this DPA," and separately commits to notify customers "without undue delay, and in any event within five (5) business days (or sooner where required by applicable Data Privacy Laws)" after discovering a security breach.

That is a real, citable risk-transfer term — the kind this index's own rubric looks for under `audit_trail` and `compliance` — and it is also narrower than it reads at first pass. The indemnity covers breaches of the DPA's own warranties or of data-privacy law; it does not extend generally to harm caused by the agent's actions inside a customer's store. And it is a contractual promise from the vendor itself, not a transfer of risk to a third party: this outlet's review of Richpanel's public materials found `insurance_available`, `insurance_carriers`, `coverage_limits` and `liability_cap` are all `no_public_information` on the current capture of richpanel.com. A buyer gets a notification deadline and a narrow indemnity; they do not get a named insurer standing behind the agent's mistakes.

Full story, with the complete DPA language and Richpanel's own compliance claims: https://aidispatch.news/articles/richpanel-dpa-indemnity-no-insurance-2026-09-28

Separately, and briefly: Munich Re aiSure — the insurance-vendor candidate whose six-field correction led last week's issue — is unchanged this week, still held at 8 present fields and still a candidate, not a published entity. The correction has held; there is nothing new to report on it.

## What changed in the index this week

Real counts, queried at draft time:

- **Published entities: 802** (771 agent, 31 vendor). Candidates: 2,416. Archived: 2. - **4 candidate-to-published promotions, 5 published-to-candidate demotions** (net −1) in the trailing seven days, per `entity_overwrite_log`. - **79 `entity_fields` rows written as new or corrected `present` values** in the trailing seven days, against **378 flipped to `no_public_information`** in the same window — a far larger share toward absence than new claims. This newsletter reports the count without speculating on cause; it is a fact about this index's extraction and review pipeline, not about the published index's accuracy. - **`change_queue`: 0 new items logged in the trailing seven days** — the newest detection row in the table is dated 2026-09-24, eight days before this issue. 285 items were resolved against the existing backlog in the same window; the pending backlog now stands at **3,873**. - **Zero published entities carry the withdrawn NSFW subtype** — the publication-floor and NSFW guards both held clean this week.

## Sponsorship

No listing is sold in this issue. A direct query of `listings` for `tier = 'featured' AND status = 'active'` returned zero rows at draft time. There is nothing to disclose because nothing has been sold.

## Methodology

Full rubric and sourcing standard: https://aidispatch.news/methodology. Every count above was queried directly against the live database at draft time rather than carried forward from a prior issue.

---

*AI Dispatch is not a growth-hacked directory. Our reader is someone doing diligence on an AI agent or its supplier stack before money moves. We publish absence as a finding, we cite everything, and this issue's assurance development is a vendor adding contractual risk-sharing language that still falls well short of insurance — which is exactly the gap this index exists to measure. If we ever stop doing that, stop trusting the newsletter.*