S1 · AI Landscape Quickstart
A fast, source-linked orientation to the AI landscape and Anthropic’s place in it. Use this to open briefings and to set shared vocabulary.
How the Anthropic Counsel pack fits together
This page is your map. It ties together the client brief (S1–S6), Foreseeable Misuse, the Penumbral Privacy Spine, policy overlays, the reading stack, and the tools & templates.
Client Brief (S1–S6)
Use S1–S6 when you are telling the high-level story to boards, executives, and regulators. Each segment has a specific job; together they form a coherent counsel narrative.
S2–S3 · Who’s Who & Distinctives
Map the key actors in the ecosystem and highlight Anthropic’s distinctive governance and technical posture (ASL-3, RSP, transparency work).
S4–S6 · Why our recs & acceptance
Connect Anthropic’s posture to concrete recommendations and the acceptance matrix. Use these segments to structure decisions and record trade-offs.
Bundles around the brief
These bundles extend S1–S6 into scenarios, doctrine, and supporting material. Treat them as overlays on top of Anthropic’s live policies and technical reports.
Foreseeable Misuse pack
Scenario library and acceptance table for risky uses of Claude. Helps you decide what to accept (with controls), defer, or decline.
Penumbral Privacy Spine
Organises key privacy and autonomy cases into interpretive packets, linked to an acceptance table and counsel usage guide.
Policies, tools, reading stack & welfare
Use the policy overlays, tools & templates, reading stack, and welfare handout as your reference shelf when turning S1–S6 into concrete client work.
Policies & overlays · Tools & templates · Reading stack · Model evals & welfare
How to use this pack in live matters
Treat these pages as a map and a set of overlays, not as an auto-pilot. Use the steps below when you turn S1–S6, Foreseeable Misuse, the Penumbral module, or the welfare handout into live legal work product.
1. Start from the right bundle
Begin here, then click into the bundle that matches your task: S1–S6 for the story, Foreseeable Misuse for risky scenarios, the Penumbral module for doctrine, the welfare handout for model behaviour, or the tools/templates when you are ready to structure a decision.
2. Pull the RPE watchpoints
Each high-stakes bundle is wired to an RPE watchpoints JSON file. Skim the relevant bullet points
(for example, client_brief_rpe_watchpoints_v0_1.json, fm_rpe_watchpoints_v0_1.json,
welfare_rpe_watchpoints_v0_1.json, penumbral_rpe_watchpoints_v0_1.json) before
you rely on that bundle in opinions, board packs, or redlines.
3. Anchor every move to a source
Use the Policies & Overlays and Reading Stack indexes to find the live RSP, ASL-3 docs, evaluation reports, and key cases. Cite those directly in your work product, including version or date, rather than treating this pack as the primary source.
4. Check staleness and overreach
Ask what would have to change in Anthropic’s policies, product behaviour, or law for your advice to become misleading. Pay special attention when you are making claims about AI welfare, rights, or penumbral privacy holdings; keep those tethered to the underlying texts and uncertainty notes.
5. Record the decision
When you make an accept/mitigate/defer/decline call, use the S5/S6 tools to capture the decision, the upstream sources and bundle pages you relied on, and any RPE watchpoints that were especially salient. Treat that record as part of your institutional memory.
RPE watchpoints for the client brief
The Risk-Prediction Engine (RPE) runs in the background for this project. For S1–S6, it surfaces a small set of watchpoints you should keep in view when you rely on the brief in opinions, board packs, or redlines.
What to watch for when using S1–S6
- ASL-3 / RSP are not guarantees. The brief should not be treated as creating hard legal guarantees or strict liability shields. Always pair it with live RSP and ASL-3 documents and client-specific analysis.
- Do not treat S1–S6 as the only source. Use the Policies & Overlays index and Reading Stack to route back to the underlying terms, policies, and evaluation reports before copying language.
- Guard against version drift. When you reuse S1–S6-derived language, label which policy and model versions you are relying on and check for newer Anthropic materials.
- Record how you supplemented the brief. In opinions and board papers, note which additional contractual, regulatory, or technical controls you are counting on beyond Anthropic’s internal posture.