IntegrAuth Academy
Identity, security and practical AI — free, from zero. Byte-sized lessons, real diagrams, hands-on labs, zero fluff. Two programs, each with its own exam and certificate.
English is hard? Your browser can translate this page into Tamil, Telugu, Kannada, Malayalam or Hindi. Tech words stay in English so you learn the real names.
How?
- Chrome on phone: ⋮ menu → Translate
- Chrome on computer: right-click → Translate to …
- Edge: the translate icon in the address bar
Cheat sheet & pop quiz
You finished the track. Here it is on one page, plus 5 quick questions. Your goal: answer each one before you reveal it.
If you remember nothing else
| Lesson | The idea to keep |
|---|---|
| Your first AI chat | An assistant is an app around a model. Each chat is its own context. A shared link is a public page. Never paste passwords or other people's details. |
| Where AI is strong, and where it fails | The frontier is jagged. Wrong answers sound just as sure. Ask "what if it's wrong?" and "could I tell?" Then pick a step on the trust ladder. |
| Everyday wins with AI | Give a goal, context, audience, format and example. The first answer is a draft. |
| Research and fact-checking with AI | A citation is a claim, not proof. Open it, find the sentence, check the date and read laterally. Trace it to the primary source. A forward with a hurry, a link or an OTP request is a warning sign. |
| Health, money and legal questions | AI can explain and help you prepare. A qualified person decides. Stop at a dose, a deadline, a sure profit or “you don’t need a lawyer”. |
| Images, voice and video | Change one thing at a time. Get consent for real people. Label realistic fakes. |
| AI in your own language | Answers are often weaker outside English. Read back voice typing. Check key facts, and check important translations with a person. |
| Get real answers from your files | The AI reads a converted copy of your file. Numbers without code were written, not calculated. Ask for quotes and check 2 figures. |
| Tech basics: terminal, JSON and keys | The terminal does exactly what you type. JSON is strict. Keep API keys in an environment variable or a git-ignored .env. Revoke a leaked key. Commit before an AI changes your files. |
Pop quiz: 5 questions
Q1 · Priya planned a trip in one chat. The next day, she opens a new chat. She types "Now make day two cheaper." The assistant asks which trip she means. Is something broken? What should she do?
Nothing is broken. Each chat is its own context. A new chat sees only its own messages, plus any custom instructions and saved memories. It does not see the trip chat. She should go back to the trip chat, or paste the plan into the new one. See your first AI chat.
Q2 · Priya uploads a sales file with 1,240 rows. She asks for each region's total. The answer looks neat and shows no code. It mentions "the 1,000 rows provided." What should she think?
There are 2 warning signs. No code means the totals were written, not calculated. And "1,000 rows" means part of the file was cut. She should ask it to run code on the whole file and show the method. Then she checks one region's total herself. See get real answers from your files.
Q3 · An AI answer about new recycling rules links to "the Greenfield Recycling Council." The page loads and says exactly what the answer says. Priya has never heard of this council. Is she done checking?
Not yet. The link exists and supports the claim. But she has not judged the source. She should read laterally. What do other sites say about this council? Who pays for it? She should check the date. Best of all, she finds the rules on the official government site. That is the primary source. See research and fact-checking.
Q4 · A colleague is making a farewell video for a retiring manager. He wants to copy her voice with AI, from old meeting recordings. He plans to keep it a surprise. What would you suggest?
Don't make the copy without asking. A voice clone of a real person needs that person's consent. A surprise rules that out. Many good services only let people clone their own voice. Ask the manager first. Or use a stock AI voice or real recorded messages, and label any AI voice. See images, voice and video.
Q5 · Priya pasted her API key into a code file. She uploaded ("pushed") it to a public repository, an open online code folder. 10 minutes later, she deleted the line and pushed again. Is the key safe now?
No. Git keeps every earlier commit, so the key is still in the history. Bots scan public code for keys. She must revoke the key in her account on the company's website. Then she makes a new one. She keeps the new key in an environment variable or a .env file listed in .gitignore. See tech basics.
Go deeper: a checklist before you trust an answer
- If this is wrong, what happens? Could I tell? See the trust ladder.
- Is it a health, money or legal decision? Then a qualified person decides. See health, money and legal questions.
- Did it have what it needed: the right chat, my context, the real file? See the 5-part recipe.
- Did it search, run code, or answer from memory? See research and fact-checking.
- Did I open the source, or check the number myself? See get real answers from your files.
- Did I share anything I shouldn't, like secrets or someone else's face or voice? See safe AI at work.
- Do I understand the command it wants to run? Is my work committed? See tech basics.
That's AI from zero. You can talk to any assistant and match your checking to the stakes. You can check a citation, a number and a file. You can make images and voices with care. You can read the commands AI tools use. Next up — AI essentials: why AI behaves the way it does. Start at the AI essentials start page.
Pick a real task from your week. Ask any free assistant, using the 5-part recipe. Check one fact at its source. Then open the settings and look at your memory and shared links.