Your command tools: the Briefcase, the professor, the dossier you will build
The short answer
Three tools, carried all program
The Briefcase (prompts you have tested and can vouch for), the professor (an examiner that interrogates your work), and the dossier (an evidence file that grows topic by topic) are the instruments you use in every remaining topic. Set them up in Module 0; use them to the capstone.
What you will be able to do
- Name the three command tools you use throughout this program (the Briefcase, the professor, and the dossier) and state in one sentence what each one is for.
- Explain why a disciplined AI toolkit and an evidence trail are governance instruments, not conveniences, using the Sports Illustrated failure as the anchor case.
- Set up an evidence dossier for your own organization, with a running Build Log index, and file your first artifact into it (the AI systems inventory you built in Topic 0.2).
- Seed your AI Briefcase with your first vetted prompt, recording what it does and how it can fail, so you never paste a prompt you do not understand.
- Use the in-app professor to turn one of your own claims into a defense drill, and rehearse answering it out loud.
- Write a plain disclosure line for an artifact (what AI touched it and how you checked it), the single habit that separates you from the Sports Illustrated failure.
- Distinguish an evidence trail you can defend from a pile of outputs you merely produced, and explain why the living credential opens the first and not the second.
The ideas, one by one
The Briefcase is vetted, not collected
A prompt is in your Briefcase only after you have run it and can state what it does and how it fails. Naming the failure mode is the discipline. An untested prompt is a rumor, not a tool.
The professor makes you defensible, not finished
It attacks work you have already done until you can defend it or discover you cannot. Use an assistant to draft; use the professor to be ready. When it finds a hole, fix the artifact, do not argue with the tool.
The dossier is where the proof lives
Every artifact you build is named, dated, and inspectable, indexed by a running Build Log. You cannot govern what you cannot see (the inventory), and you cannot defend what you did not record (the dossier). Both halves are command.
Disclosure is one honest sentence, recorded at the time
For every artifact AI touched, you write what AI did and how you checked it. That single line does more governance work than pages of policy, because it makes the AI use visible and names the human check.
Sports Illustrated failed on all three at once
In 2023, the magazine published reviews under fabricated AI author profiles with no disciplined tooling, no defensible account, and no evidence trail (Futurism, November 2023). The failure was not the AI; it was the absence of the three tools. Concealment, not technology, made it indefensible.
Start now because evidence has a short honesty half-life
A dossier fed from Module 0 is honest, recorded when things happened. A dossier assembled at the end is reconstructed from memory, which is fragile and dishonest. The capstone audit finds a trail only if you started one early.
The living credential opens the dossier, it does not claim knowledge
An employer inspects your real artifacts and defenses, not an adjective on a certificate. The credential is only as strong as the dossier it opens, and the dossier is only as strong as the discipline you start today.
The tools embody the program's laws
The Briefcase is "build before you govern," the professor is "adversarial by default," and the dossier is "consequence over coverage." Setting them up is not overhead; it is the discipline in miniature.
You run the tools; the tools do not run you
A vetted prompt still needs a human read. A defensible artifact still needs a human signature. The tools make command possible; they never replace the commander.
A good dossier is boringly simple
One place you control, one Build Log that indexes everything, one file per artifact with a name and a date. The common failure is not plainness but splintering across email, chat, and personal drives; the discipline is convergence, not decoration.
The tools are domain-independent because the failure is
The Sports Illustrated structure (undisciplined AI use, no defense, no trail) can happen in logistics, insurance, or healthcare as easily as in publishing. If your organization ships anything AI helped produce, the same three questions apply and the same absence of tools produces the same indefensible position.
Listen to it
The same lesson, talked through by two hosts as episode 3 of the podcast.
Real cases
These examples show the presence or absence of the three tools in real cases, with the reasoning stated plainly. Each is real and cited. Except for the Sports Illustrated anchor, which this topic owns, other events are pointer-referenced to the topics that treat them in depth.
Example 1: Sports Illustrated and the fake AI authors (the anchor). In November 2023, Futurism reported that Sports Illustrated had published product reviews under fabricated author profiles, including "Drew Ortiz," whose headshot was sold on a site offering AI-generated portraits (Futurism, "Sports Illustrated Published Articles by Fake, AI-Generated Writers," November 2023). The content came from a third-party firm, AdVon Commerce. The Arena Group, Sports Illustrated's publisher, said it ended the partnership, removed the content, and opened an internal investigation, and stated that AdVon had assured it the articles were written and edited by humans; sources familiar with the work disputed that account (Variety and PBS NewsHour, November 2023).
Read through the three-tool lens: there was no Briefcase (AI use was undisciplined and outsourced), no honest defense available (no one could answer how the work was made), and no dossier (no trail, no disclosure, nothing to inspect). The failure was not the AI. It was the absence of all three tools at once.
Example 2: A newsroom that disclosed and survived (contrast). Many reputable publishers now run AI-assisted work openly, with a stated policy on where AI is used and a human editor accountable for every piece. The contrast with Sports Illustrated is instructive: the technology can be identical; the outcome diverges entirely on whether the use is disclosed and recorded.
A published AI-use policy is a Briefcase habit made organizational; a named human editor is the human check the disclosure line records. This is the mundane, unglamorous discipline that keeps AI use from becoming a scandal. (General industry pattern, 2023 to 2026; verify any specific outlet's current policy before citing it.)
Example 3: The AI systems inventory as a dossier's first entry. The discipline of writing down every AI system an organization runs, with an owner and a purpose, is the "you cannot govern what you cannot see" move that Topic 0.2 owns. (see Topic 0.2) For this topic, the point is what happens next: that inventory becomes entry 001 in your dossier, indexed in your Build Log, with a disclosure line if AI helped you build it.
An inventory that lives in one person's head or a lost spreadsheet is not evidence. The same inventory, filed and indexed, is the first inspectable artifact an auditor can open.
Example 4: The disclosure letter after an incident. When an AI feature causes harm, the decision to tell the customer, and exactly what to say, is a governance act with real consequences. Module 3 owns the deep treatment of the customer disclosure decision. (see Topic 3.7)
Here it illustrates the dossier principle in its highest-stakes form: the letter you send, and the record of when and why you sent it, becomes a permanent dossier artifact. An organization that discloses cleanly and keeps the record is defensible; one that conceals, as in the anchor case, is not. The disclosure habit you start in this topic is the small daily version of that same discipline.
Example 5: The capstone audit finds a trail, not a scramble. In the capstone, a seven-AI board and a live examiner inspect the dossier you have accumulated across the whole program. (see Topic 13.1) (see Topic 13.3) The learners who set up their dossier in Module 0 and fed it every topic walk in with a navigable evidence file. The learners who did not spend the capstone reconstructing from memory, which is exactly the position Sports Illustrated was in when Futurism called.
The example is you, twelve modules from now. Which position you are in is decided by the habit you start today.
Example 6: The AI-authored content problem beyond one magazine. The Sports Illustrated case is not an isolated scandal. It is a pattern that has recurred wherever AI-generated content ships without disclosure or a human check. Public reporting has documented AI-invented book titles listed beside real authors in a syndicated newspaper reading guide, and AI-written how-to guides sold under fabricated author names on major retail platforms, some of them offering unsafe advice before being removed (NBC News and Fortune, 2023 to 2025; these events are treated as pattern context here, not as centerpieces owned by this topic). The lesson transfers: the failure structure repeats across industries and formats, undisciplined AI use, no disclosure, no trail. The three tools are the general defense precisely because the failure is general. If your organization ships anything an AI helped produce, whether content, a routing decision, a customer notice, or a risk score, the same three questions apply, and the same absence of tools produces the same indefensible position.
The pattern is not limited to publishing. A public-sector exam algorithm graded students by their school's past performance instead of their own work, with no defensible account ready when the public objected, and the government scrapped it within days. (see Topic 8.3) A government welfare system ran dozens of AI fraud-detection models with too little visibility into how they scored people, which a human-rights investigation later documented in detail. (see Topic 5.3) Different sectors, different stakes, the same structural absence: no disciplined tooling, no rehearsed defense, no trail an outsider could inspect.
Example 7: The provenance stamp that answered a question years later. Consider the ordinary case that never becomes a headline precisely because the tools worked. An operations lead files a vendor risk summary in her dossier with a provenance stamp: AI drafted it, she verified the retention and access claims against the signed contract, she is the named signer, and one claim about the vendor's subprocessors is flagged UNRECORDED pending confirmation.
Two years later, after a data incident at that vendor, an auditor asks what her organization knew and checked at onboarding. She opens one file. The answer is complete, dated, and honest, including the one thing she had not confirmed, which shows diligence rather than concealment. There is no scandal here, no story to cite, which is the point: the provenance stamp turned a potential crisis into a two-minute retrieval. The tools are most valuable in exactly the cases you never hear about.
Example 8: The living credential an employer could actually inspect. A hiring manager reviewing two candidates sees that both list AI governance training. One holds a conventional certificate asserting completion of an exam. The other holds a living credential that, when opened, reveals a dossier: a real AI systems inventory, a conformity file that survived a documented attack, an incident log with the disclosure letter that followed, and a viva transcript.
The manager cannot inspect what the first candidate actually did; they can inspect exactly what the second did, decision by decision. This is the payoff of the dossier discipline started in this topic, viewed from the other side of the table: the evidence trail is not only your defense in an audit but your proof to an employer, and it is inspectable precisely because every artifact stayed nameable from the day it was made. (see Topic 13.1)
Where people go wrong
- "The Briefcase is just a folder of prompts I found online." No. A prompt is in your Briefcase only after you have run it and can state what it does and how it fails. An untested prompt is a rumor, not a tool. The whole value of the Briefcase is that you can vouch for everything in it; a folder of unvetted prompts has none of that value and all of the risk.
- "The professor is there to give me the answers." The professor is the opposite of an answer machine. It interrogates work you have already done and pushes until you can defend it or discover you cannot. If you use it to write your artifacts, you have skipped the point and built nothing defensible. Use an assistant to draft; use the professor to be ready.
- "I will start my dossier when I have something worth putting in it." You already do. Your Topic 0.2 inventory is a real artifact and belongs in the dossier now. Waiting until later means reconstructing a trail from memory, which is dishonest and fragile, and it is precisely the position that made Sports Illustrated indefensible. The dossier's value comes from being started early and fed continuously.
- "Disclosure is a legal disclaimer I paste at the bottom." A disclosure line is a plain fact recorded at the time: what AI did and how you checked it. It is not boilerplate and it is not an apology. A single honest sentence ("AI drafted this; I verified the names and fixed four errors") does real governance work that a generic disclaimer does not, because it names the human check.
- "A vetted prompt means I do not have to read the output." Vetting lowers the odds of a bad output; it never removes your responsibility to read what the tool produced before it reaches anyone. The human check is the point of the disclosure line. A Briefcase makes AI use deliberate; it does not make it unsupervised.
- "The dossier is for the capstone; it does not matter until then." The dossier is used the entire program, and its whole strength is continuity. The capstone audit and the living credential both open the dossier you have been keeping, not one you produce at the end. Treating it as a last-minute task guarantees the scramble the program is built to prevent.
- "The professor can be wrong, so I should not trust it." The professor can be wrong at the edges, like any AI tool, which is exactly why you verify its factual claims against primary sources before they enter an artifact. That does not make it untrustworthy for its actual job, which is to attack your reasoning and expose weak defenses. Use it for interrogation; verify it for facts.
- "These tools are overhead that slows down real work." They are the fast path, not the slow one. Benjamin set up all three in under an hour and immediately found a real gap that could have become a public failure. The slow path is the Sports Illustrated path: move fast, record nothing, and spend months on damage control when a stranger asks one question you cannot answer.
- "If I use good AI tools, my governance is handled." Tools do not govern; you do. Sports Illustrated's failure was not a bad AI tool; it was the absence of discipline, defense, and evidence around whatever tools were used. The Briefcase, the professor, and the dossier only work because a human runs them deliberately. The tools make command possible; they do not replace the commander.
- "A prompt I vetted once is vetted forever." A prompt's behavior can change when the model behind it changes, so a prompt vetted against one version of a tool may fail quietly after the vendor updates it. The Briefcase is a living kit that needs periodic re-testing, not a museum. A card that has not been re-confirmed in a long time should be treated with the same suspicion as an untested prompt, especially after any known change to the underlying model.
- "My organization is too small for a dossier." Size is not the test. AI use is. A two-person company that ships AI-assisted work to customers can be asked exactly the same question Sports Illustrated could not answer, and a small organization often has less margin to absorb a public failure. The dossier scales down cleanly: for a small operation it may be a single folder with a handful of files, which is still infinitely more defensible than nothing recorded at all.
- "The professor and my own AI assistant are the same thing, so I only need one." They do different jobs. Your own AI assistant produces drafts; the professor attacks them. Using only an assistant means your work is never adversarially tested before a real examiner sees it; using only the professor means you have nothing to draft with. The design intent is to draft with an assistant, harden with the professor, and record the result in the dossier. Collapsing the two roles loses the adversarial step that makes the work defensible.
Questions people ask
- What is briefcase (AI Briefcase)?
- A small, growing collection of AI prompts that you have personally tested and understand, each recorded with what it does and how it fails. A prompt belongs in the Briefcase only after you have run it and can vouch for it. The Briefcase makes AI use deliberate rather than accidental and is fed by the ready-to-use prompts in every topic of this program.
- What is professor?
- The in-app examiner in this program that interrogates work you have already done, drills you, and rehearses your defense. It does not produce your artifacts; it attacks them until you can defend them or discover you cannot. It is grounded in the program's material and can still be wrong on facts, so its factual claims are verified against primary sources before entering an artifact.
- What is dossier?
- The accumulating evidence file where every artifact you build in this program is named, dated, and inspectable. It grows from Module 0 to the capstone, is indexed by a Build Log, becomes a formal evidence annex in Module 10, and is what the living credential opens and the capstone audit inspects. More on Dossier
- What is Build Log?
- The running index of the dossier, holding one line per artifact: a number, what it is, the topic it came from, the date, and where it lives. The Build Log keeps the dossier navigable and inspectable as it grows; without it, a large dossier is a haystack rather than a record. More on Build Log
- What is artifact?
- Anything you produce in this program that an employer or auditor can open and inspect: an AI systems inventory, a priorities memo, an eval suite, a conformity file, an incident log, an investment memo. Every artifact belongs in the dossier, and every assessment in this program is an artifact plus your defense of it.
Keep going
This lesson builds AI inventory and use-case intake, and that page shows the roles that hire for it. Every Certified AI Governance Professional (CAIGP) lesson.
Watch the lesson and prove it
Proving it happens inside the program: watch, explain it in your own words, and it goes on a record employers can check.
Watch the lesson and prove it →