Earn the Open Certificate in AI Literacy
The Level 100 credential of How AI Works. It is free. It is earned, never bought: paying cannot replace a host's approval (credential spec).
This page shows the whole route: what you do, how your project is marked, what happens if it is "not yet", how to appeal, and exactly what the certificate does and does not claim. Anything marked Proposal is not decided yet. Bob Rapp and Ken Johnston, the review board co-chairs, decide it unless the page names Bob alone. No certificate has been issued yet.
What you must do
- Take the placement test if you like. It suggests where to start and which gear to read in.
- Work through the Level 100 modules in plain, standard or technical gear. All three gears teach the same objectives and share one quiz.
- Do each module's quiz and workbook. They are practice. Your quiz answers stay in your browser, and nothing about them goes on the certificate.
- Do the project, 100.P: your rule and receipt. A second person reviews it, then a host marks it on the Foundation's Thursday call.
- When a host marks your project as passed, you can receive the certificate as a signed record.
Proposal, needs the review board co-chairs’ (Bob and Ken) decision: quizzes and workbooks stay practice only and are not part of the pass decision, because nothing about them is recorded. Today the credential spec says to "complete" them; this proposal would make that wording a recommendation, not a requirement.
The project in one line
Bring a rule in your own words for one AI action in your job or your daily life, with a receipt (prompt, answer, check, decision, owner) and a recovery plan. A second person reviews it, and a host marks it on the Foundation's Thursday call. That is objective 100.P-a in the Level 100 objectives map.
The module walks you through it step by step: Project: your rule and receipt.
The rubric: how your project is marked
Rubric version 0.1, draft. A host marks each of the four parts as Meets or Not yet.
| Part | Meets when | Not yet looks like |
|---|---|---|
| Rule | All six parts are filled in, a coworker could follow it without asking, and it says what happens in the two "otherwise" cases | A part is missing or vague ("when it is safe"); a coworker would have to ask you what it means; or one or both "otherwise" cases are missing or do not say who decides |
| Receipt | If run: it shows the prompt, the answer, the check, the before and after decision, and names the owner. If held or denied: it shows the proposed action, "not run" for the prompt and answer, the check that led to the decision, the before decision and why, the "otherwise" followed, and the owner. Either way, private details are removed | A part is missing; the decision was written after the AI ran instead of before; a held or denied receipt has a made-up answer; the owner is "the team" instead of a named person; or the shareable copy still has a name, account number or other detail that points to a person or to confidential work |
| Check | If run: every fact, number, name or date that was used was checked against something outside the AI, and anything not checked is marked "not checked" and was not used. If held or denied: it says what was checked to reach the decision and what is still unknown | Something that was used was not checked; the "check" was asking the same AI again; unchecked parts are not marked; or a held or denied receipt does not say what was checked or what is still unknown |
| Recovery | It names the first step, the person to tell and the fix | The person to tell is "someone" or "my manager" with no name; the first step is not something you could do today; or there is no fix for the people affected and no change to the rule |
The "Meets" column is the host's table from the 100.P module, word for word. The "Not yet" column is new. Proposal, needs the review board co-chairs’ (Bob and Ken) decision: adopt these "Not yet" descriptors as rubric version 0.1.
The pass rule. Your project passes when all four parts meet. There are no points and no partial pass: a strong receipt cannot make up for a missing recovery plan. A held or denied project can pass. Nobody has to run an action their rule says no to. Proposal, needs the review board co-chairs’ (Bob and Ken) decision: this all-four rule, already in the 100.P module, becomes the certificate's pass rule. The reasons are in the assessment dossier.
A worked example: Maya's first try, and her second
Maya is a shift supervisor in the paint shop at Big Widget Company. She and the company are made up. Her finished project is the worked example in the 100.P module. Here is a made-up first try, so you can see how a host marks a project that is not there yet.
First try. Her rule says the AI may summarize line-stoppage notes "when the notes are clean." It has one "otherwise" case. Her receipt is complete and has no names in it. Her check says she counted one of the three causes in the raw log. She used all three counts in the shift meeting. Her recovery plan says "stop and tell someone."
| Part | Host's mark | What the host wrote |
|---|---|---|
| Rule | Not yet | "Clean" is not something a coworker could check. Say what clean means: the company-approved tool, notes the policy allows, names removed. Add the second "otherwise" case: notes you cannot tell are covered. |
| Receipt | Meets | All five parts are there, the decision came before the run, and nothing points to a person. |
| Check | Not yet | You used three counts and checked one. Count the other two in the raw log, or mark them "not checked" and do not use them. |
| Recovery | Not yet | Name the person to tell, and say how you would change the rule. |
Second try. Maya defines "clean", adds the second case (she holds and asks her manager), counts all three causes in the raw log, finds one count was too high and fixes it, and names the maintenance lead in her recovery plan. All four parts meet, and the project passes. That second try is the finished example in the module.
How a host marks it
These steps come from the 100.P module and the Foundation's Level 100 Begin worksheet.
- Bring it to the Thursday call. As of November 2026, the worksheet says the call is on Thursday at 15:00 Pacific and the first five minutes are for receipts. You show your work to one person and say your rule out loud.
- Cannot make a Thursday? The worksheet lets you send your work so a host can mark it later.
- No computer? Do the writing on paper. A host lends a laptop on the call.
- Share only the shareable copy, with private details removed.
- A person marks it. The host is a person. The AI decides nothing, signs nothing and marks nothing.
Proposal, needs the review board co-chairs’ (Bob and Ken) decision: the host writes down a mark for each of the four parts, the rubric version and the date, and gives you one next step for each "Not yet". A host does not mark a project they reviewed as your second person, and does not mark the project of someone they manage.
If it is "not yet": revise and resubmit
"Not yet" is not a fail. Fix the parts the host named and bring it back. Paying cannot replace approval.
Proposal, needs the review board co-chairs’ (Bob and Ken) decision: there is no limit on tries and no waiting time. Bring it back on any later Thursday, or send it. The host marks all four parts again, because fixing one part can change another.
Appeals
Proposal, needs the review board co-chairs’ (Bob and Ken) decision: if you think a part was marked wrongly, you can ask for a second look within 30 days of the mark. A different host, who did not mark it the first time, marks all four parts again against the same rubric version within 14 days. If the two hosts disagree, the co-chairs decide, or name a board member who has not marked the project to decide. You can always revise and resubmit instead of appealing.
Needs Bob's decision: how you ask for a second look. There is no appeal contact yet, so none is given here.
If you think a lesson itself is wrong, that is not an appeal. It is a correction, and each module keeps a public errata log.
What the certificate says, and what it does not
It says: the name you choose, "Open Certificate in AI Literacy", Level 100, the issue date, your public project title, the host who approved your project, a receipt ID and a link to check it. It carries the lines "Free and open. Every source cited. Human-reviewed." and "Earned, never bought." The signatories are Bob Rapp and Ken Johnston, each titled “Founder and Co-chair, Review Board”.
It means: a named person marked your Level 100 project as meeting all four parts of the rubric, on that date.
It does not mean:
- It is not a degree, a "Master of" anything, a university award or an accredited qualification.
- It is not a professional certification or a vendor certification.
- It is not a legal approval, and it does not show that your company's use of AI meets any law.
- It does not show a score on every lesson. Quizzes are practice and are not recorded.
- It does not check who you are. You choose the name on it; there is no legal-name check.
- The project was open-book and not supervised, and a second person reviewed it with you.
- It is not the Foundation's Practice 100. Practice 100 is a separate mark that also needs the worksheet's signed Beacon bundle. Needs Bob's decision: whether one host mark can count for both.
What the record holds. Only the fields above and what is needed to check the signature and its current status. No quiz answers, no project text and no email.
What a valid signature shows. That the record has not been changed since the Foundation's key signed it, and whether it is still valid, revoked or replaced. It does not show anything more than the claim above. A picture of a certificate proves nothing on its own; the check does.
A sample record (TEST KEY)
This sample is signed with the course's public TEST key. Anyone with the repository can sign with that key, so it proves nothing. It is not a credential, and the learner and host are made up.
The course site builds the sample from the course's credential code, so it always matches the record format the verifier checks.
TEST KEY · Sample only · Not a real credential
| Credential | Open Certificate in AI Literacy |
|---|---|
| Display name | Maya (sample learner, made up) |
| Project title | Paint-shop stoppage summary: rule and receipt (sample) |
| Reviewing host | Sample host (made up) |
| Issue date | 2026-11-20 |
| Level | 100 |
| Receipt ID | urn:uuid:00000000-0000-4000-8000-000000000100 |
| Issuer | AiGovOps Foundation — TEST ONLY |
The signed record (JSON)
{
"@context": [
"https://www.w3.org/ns/credentials/v2",
"https://purl.imsglobal.org/spec/ob/v3p0/context-3.0.3.json",
{
"@vocab": "https://verify.aigovops.org/vocab#",
"projectTitle": "https://verify.aigovops.org/vocab#projectTitle",
"reviewingHost": "https://verify.aigovops.org/vocab#reviewingHost",
"issueDate": "https://schema.org/dateIssued",
"level": "https://verify.aigovops.org/vocab#level"
}
],
"id": "urn:uuid:00000000-0000-4000-8000-000000000100",
"type": [
"VerifiableCredential",
"OpenBadgeCredential"
],
"name": "Open Certificate in AI Literacy",
"issuer": {
"id": "https://aigovops-foundation.github.io/how-ai-works/verify/.well-known/issuer.json",
"type": "Profile",
"name": "AiGovOps Foundation — TEST ONLY"
},
"validFrom": "2026-11-20T00:00:00Z",
"testKey": true,
"credentialSubject": {
"id": "urn:uuid:00000000-0000-4000-8000-000000000100#subject",
"type": "AchievementSubject",
"name": "Maya (sample learner, made up)",
"projectTitle": "Paint-shop stoppage summary: rule and receipt (sample)",
"reviewingHost": "Sample host (made up)",
"issueDate": "2026-11-20",
"achievement": {
"id": "https://aigovops-foundation.github.io/how-ai-works/verify/achievements/100",
"type": "Achievement",
"name": "Open Certificate in AI Literacy",
"level": 100,
"description": "Free and open. Every source cited. Human-reviewed.",
"criteria": {
"type": "Criteria",
"narrative": "Complete the level modules, each five-question quiz and workbook, and a project approved by a human host. Earned, never bought."
}
}
},
"credentialStatus": {
"id": "https://aigovops-foundation.github.io/how-ai-works/verify/status/1.json#1",
"type": "BitstringStatusListEntry",
"statusPurpose": "revocation",
"statusListIndex": "1",
"statusListCredential": "https://aigovops-foundation.github.io/how-ai-works/verify/status/1.json"
},
"proof": {
"type": "DataIntegrityProof",
"cryptosuite": "eddsa-rdfc-2022",
"created": "2026-11-20T00:00:00Z",
"verificationMethod": "https://aigovops-foundation.github.io/how-ai-works/verify/.well-known/issuer.json#key-1",
"proofPurpose": "assertionMethod",
"proofValue": "zVLLQ8V44BoK3ih8Dgc245m9sHVnh3WZjPksTpQUDk83TqpdUgkG47sPyghDxajuKPVzrUhGYTXxFJmoQQoj3tXf"
}
}The evidence behind this route
- Assessment dossier for Level 100: objective to item to rubric, marker calibration, and why the pass rule is all four parts.
- Credential names, learner process and record.
- The truth standard every lesson passed before publication.
Start here
The project is the last module. If you have not read the lessons yet, start with 100.1.