Velaru Standards · L1 Process Physics · v1.0

ACCESS-1 — Access Trace Receipt Profile

Normative receipt profile externalizing access-trace integrity for governed agent episodes: workspace admission, global broadcast, downstream consumption, metacognitive assessment, and human oversight — cryptographically bound to a frozen Handler Receipt.

Normative non-claim: ACCESS-1 certifies operational access-trace artifacts derived from consciousness-science indicator properties (Butlin et al., TICS 2025). It does not certify phenomenal consciousness, welfare status, or model correctness.

Deliberation Atom GWT operational mapping CWI process layer DRP-1 STIT companion Handler Receipt binding

Problem statement

Regulators under EU AI Act Articles 12–14 require decision-level traceability and meaningful human oversight. Consciousness science distinguishes access artifacts (what was globally available for action) from phenomenal experience (unsettled for AI systems as of 2026). ACCESS-1 implements the governable subset: externally verifiable access trace under committed policy.

Three-layer verification hierarchy

Layer	Construct	Velaru artifact
1	Continuity Receipt — what happened	L0 hash chain · Ed25519 · /verify
2	Workspace Integrity — how it happened	ACCESS-1 · CWI dimensions
3	Cognitive Emergence Standard — what it would mean	Out of product scope (informative only)

Normative claim set

workspace_admission — competing representations + selection criterion (GWT-1, GWT-2)
broadcast — selected representation digest + broadcast scope (GWT-3)
consumer_bindings[] — downstream module consumption (GWT-4)
metacognitive_assessment — reliability, uncertainty, dissent preservation (HOT-2)
human_oversight — authorized_by, oversight mode (Art. 14)
handler_receipt_id — pre-incident policy commitment (TCB)
access1_binding_hash — independent recomputation at verify

access1_binding_hash = SHA-256(profile_version ∥ episode_id ∥ handler_receipt_id ∥ competitors_digest ∥ broadcast_digest ∥ consumer_digest ∥ metacognitive_digest ∥ session_id ∥ domain)

Deliberation Atom API

POST https://velaru.onrender.com/tcb/deliberation-atom
Content-Type: application/json

{
  "selected_representation": "BIND_ALLOW: policy EPL-2026-08 satisfied",
  "competitors": [
    {"module": "classifier", "salience": 0.91, "payload": {"label": "SAFE"}},
    {"module": "policy_engine", "salience": 0.87, "payload": {"clause": "4.2"}}
  ],
  "selection_criterion": "policy_weighted",
  "broadcast_scope": ["belief", "plan", "output"],
  "consumer_bindings": [
    {"module": "tool:bind-check", "action": "invoked"},
    {"module": "carrier:exhibit-d", "action": "exported"}
  ],
  "metacognitive_assessment": {
    "uncertainty": 0.06,
    "reliability_flag": "high",
    "dissent_preserved": true
  },
  "authorized_by": {
    "principal": "underwriter@carrier.example",
    "method": "oidc",
    "subject": "auth0|..."
  },
  "session_id": "pilot-1",
  "domain": "hiring"
}

→ 200 { "entry_id", "episode_id", "access1_binding_hash", "theory_indicators",
         "handler_receipt_id", "verify_url", "deontic", "signature" }

GET https://velaru.onrender.com/tcb/deliberation-atom/{entry_id}/verify.json

Theoretical basis (informative)

Butlin, P., Long, R., et al. (2025). Identifying indicators of consciousness in AI systems. Trends in Cognitive Sciences. doi:10.1016/j.tics.2025.10.011
Butlin, P., Long, R., et al. (2023). Consciousness in Artificial Intelligence. arXiv:2308.08708
EM Foundation (2026). Computational Workspace Integrity
Mikeda, A. (2026). Precautionary framework for consciousness uncertainty. AAAI Symposium Series.

Download normative companion draft (txt) · Indicator coverage scoreboard

Integration examples → TCB profile → Indicator scoreboard → Verify a receipt →