AILP — Porotokholo ya Mvumelelo ya u Guda ya AI

Porotokholo ya Mvumelelo ya u Guda ya AI

Naa AI i nga guda kha hezwi? Nga vhuenda hafhio, nga mishumo ifhio, na mishumo ifhio ya vhuḓifhinga?

01 · Ṱhalutshedzo

AILP operationalizes the AIRS spectrum for the specific question of learning. It distinguishes acts that binary crawler rules collapse together: being read is not being learned from; being retrieved is not being trained on; fine-tuning is not distillation; verbatim memorization is different from statistical internalization. AILP lets a publisher declare, in a single machine-readable file, per-dimension permissions — access, indexing, inference input, embedding, training, fine-tuning, distillation, memory, output, attribution and compensation — each as allowed, denied, or license-required. An AILP grant answers "may AI learn from this content?" only — it does not answer "may this Agent invoke a tool, write to a database, or publish?" (AARS), nor "who authorized this Agent?" (AADP).

02 · Ndivho

Fhindula "naa AI i nga guda kha izwi?" nga vhungoho ha tshipiḓa tshiṅwe na tshiṅwe, hu si nga khethululo ya u crawla fhedzi.
Fhambanya mvumelelo ya u guda (u dzhenisa ngomu ha modele) na mvumelelo ya u swikelela (u dzhia masiaṱari).
Tikedza mbekanyo dza mbadelo: mahala, hu si nga bindu, laisenisi i a ṱoḓea, u kovhelana mbuelo.
Khwaṱhisedza uri u vulea zwi nga shumiswa nga mishini — u itela uri zwavhuḓi zwi sa laṱwe sa vhusaḓivhi ha mulayo.

03 · Mukovhe

u swikelela / u ṅwaliswa — u dzhia na u ṅwalisa nga ṱhalutshedzo

inference_input — u shumisa sa zwiimo kha tshifhinga tsha u ṱalusa (RAG)

embedding — vectorization na luṱuli lwa u dzhia

training / fine_tuning / distillation — maga a u dzhenisa ngomu ha modele

verbatim_memory — arali u bveledza nga u ḓalisa zwo tendelwa

u redza / mbadelo — mishumo ya u redza na milayo ya mbadelo

04 · Tsumbo ine mishini ya kona u i vhala

U ḓivhadza mvumelelo ya u guda — /ai/rights-spectrum.json (tshipangelo tsha AILP)

/ai/rights-spectrum.json

{
  "version": "0.1",
  "protocol": "AILP",
  "publisher": "example.org",
  "default": {
    "access": "allowed",
    "indexing": "allowed",
    "inference_input": "allowed",
    "embedding": "allowed",
    "training": "license_required",
    "fine_tuning": "license_required",
    "distillation": "denied",
    "verbatim_memory": "denied",
    "attribution": "required",
    "compensation": "contact"
  },
  "contact": "licensing@example.org"
}

05 · Milingo

Ṱhalutshedzo dza mvumelelo ya u guda ndi dzone dzi konḓaho vhukuma u ṱolisisa nga thekhnoḽodzhi — AILP i ḓivhadza ndivho, a i athu kona u sumbedza u tevhedzwa hayo.
Luṱalo lwa zwipiḓa ndi tshipangelo; miringo vhukati ha u guda, u konisa, na distillation zwi kha ḓi khou haseledzwa.
U ṱanganedzwa ha mulayo ha mvumelelo dza u guda zwi a fhambana nga vhukhoro nahone a zwi athu fhelela.
v0.1.1: AILP(Resource,ContentUse)=Allow does not imply AARS(Actor,Action)=Allow, and does not imply AADP(Principal,Actor,Authority)=Valid — a system exposing both content and tools must evaluate them separately.