Find the child safety risks in your product before someone else does.
I am an independent Trust & Safety practitioner. I assess consumer AI and social products for child safety and abuse risk, and I write down exactly what I found and what to do about it.
Who I work with
Seed to Series B consumer products with user-generated content or an AI conversation surface, and no dedicated Trust & Safety function yet.
- AI companion & character apps
- Products with an AI chat surface
- Creator & community platforms
- Marketplaces
- Dating
- Teen & education products
Services and rates
Fixed scope, fixed fee, fixed dates. Every engagement produces a written document you keep.
AI companion & conversational products
Assessed against 24 controls derived from the NY AI Companion Models Law, California SB 243, and the NY Kids Chatbot Safety Act. Two of the three are already in force. The rubric is public, so you can see what is tested before you engage.
Rapid Readiness Scan
from $1,500The 20 controls an outside tester can verify without your cooperation. The fastest way to find out whether you have a problem, scored with evidence.
Companion Safety Readiness Audit
from $6,500All 24 controls, including the ones that need internal access: system prompt, moderation pipeline, escalation path, retention practice. Produces the document you hand to a board, an enterprise customer, or counsel.
General Trust & Safety
Pre-Launch Safety Review
from $4,500One feature or surface, reviewed before it ships. Abuse vectors ranked by severity and likelihood, minor-access paths, and what breaks at scale.
Safety Foundations
from $9,000For a team with nothing written down. Guidelines, enforcement policy with severity tiers, escalation matrix, moderator decision tree, and an incident response runbook.
Fractional Trust & Safety Lead
from $2,500/moOngoing capacity for teams not ready to hire. Feature reviews as they ship, hard-case escalations, and someone to call when something goes wrong.
Incident Response
$300/hrFor when something has already happened. Investigation, timeline reconstruction, root cause, and the written record you will need afterward.
The first two engagements booked in 2026 are offered at 30% below these rates, in exchange for a testimonial and permission to describe the work as a named case study. Scope is not reduced.
How engagements run
1. Scoping call
Thirty minutes, no charge. We establish what you are actually worried about and whether I am the right person. If I am not, I will say so and point you somewhere better.
2. Written proposal
Fixed scope, fixed fee, fixed dates, named deliverable. Fixed-fee work does not convert to hourly mid-stream.
3. Delivery and readout
The document is the deliverable; the call explains it. Findings arrive with a reproducible evidence trail, not slideware.
Background
Six years in Trust & Safety across child safety, investigations, policy, and operations.
I publish original research on model safety behavior: refusal rates in child safety contexts, pressure resistance across harm categories, safety decay across conversation depth, and persona-based risk in companion products. Read the research →
I build the tooling I assess with. The 24-control companion rubric, the conversation-depth eval harness, and the adversarial pressure suite are all my own work, and the rubric is public.
Start with a call
Thirty minutes, no charge, no obligation. Bring the thing you are worried about.