How To Set Up Claude AI Safely, Step By Step

Somebody asked me last week how to get one of these AI systems set up without stepping on a landmine. Fair question. The setup itself takes about four minutes. The parts that matter are the three or four decisions buried inside those four minutes that nobody slows down to explain.

So this is the whole walkthrough. The real website and how to know you are on it, the pricing screen and what it is doing before you notice, the privacy switch that most people misread completely, and the first four things worth typing once you are inside.

I teach real estate agents. Before that I spent 20 Yrs LAPD teaching officers. The pattern is identical in both rooms: the technology is almost never the hard part. Somebody just has to walk through it once at a normal speed, without making anybody feel slow for asking.

Step 1: Get to the real website

Open whatever browser you already use. Chrome, Safari, Edge, it does not matter.

Now the important part. Type the address into the address bar at the very top of the window, not into a search box in the middle of the page.

The address is claude.ai. C, L, A, U, D, E, dot, A, I.

Press enter and you are there.

If you do search for it instead, read the results carefully before you click. The real destination is claude.ai. You will also see claude.com, which is Anthropic's own site as well. Anthropic is the company that builds Claude, so both of those are legitimate. Scroll a little further down any search results page, though, and things start drifting. Review blogs, paid ads, and tools that borrowed the name because the name has traffic.

This is where people get taken. Not by anything sophisticated. By a page that looks close enough and asks for a credit card.

The free account never requires a card. If a page demands payment details before it will let you create a free account, you are in the wrong place. Close it, go back to the address bar, and type claude.ai yourself.

One habit worth building on day one: once you are on the real site, bookmark it. Then you never have to find it again.

Step 2: Create the account

Two doors here, and both are fine.

If you already use Gmail, click the button that says Continue with Google. Pick your account and you are in. That is the fastest path and there is no meaningful risk in it.

If you do not use Gmail, type any email address into the box and click Continue with email. Claude sends a verification message to that address. Go open your email, find the message, and click the sign in link inside it. That link brings you back and finishes the job.

You will then agree to the terms of service and the acceptable use policy.

I know almost nobody reads these. Read at least the shape of them. The acceptable use policy is the short version of what the system will and will not help with, and it is more readable than you expect. If somebody sets out to use one of these tools to hurt people, that gets noticed, and it gets handled.

While we are here, a word to my own generation. We still carry an assumption that privacy is the default and that somebody is guarding it. That assumption stopped matching reality a long time ago. Not at this signup screen specifically. Somewhere back at the phone, or the smart TV, or the twenty page agreement nobody scrolled through. Understanding what you are agreeing to is not paranoia. It is just being awake.

Step 3: The pricing screen, and what it is doing

Next comes the plan selection, and this screen deserves a slow look.

There are three tiers. Free, Pro, and Max.

Watch the Pro price. When the page loads, the annual plan is already selected, and the number it shows you is the annual rate divided by twelve. That is 200 dollars billed once for the year, which displays as roughly 17 dollars a month. The actual month to month price is 20 dollars.

I do not think that is designed to trick anybody. It is how nearly every subscription on the internet presents itself now. But you should see it happening rather than have it happen to you. Find the monthly and annual toggle and click it, so you are looking at the real number for the plan you would actually pick.

On the Max tier, worth noting, there is no annual discount at all.

Start on the free plan. For the first several weeks, free is more than enough. The free tier gives you the genuine assistant with a daily usage limit rather than a stripped down demo.

The way to think about usage: it works like quarters in an arcade game. You get a certain amount of play, and heavy work spends it faster than light work. Research, long documents, and detailed back and forth burn through it quicker than simple questions. When you start bumping the ceiling regularly, that is the tool telling you it earned the 20 dollars. Not before.

Click Get Claude for free and keep going.

Step 4: Skip the download for now

The next screen offers you the desktop app.

The desktop app is free on every plan, including the free one, so cost is not the reason to skip it. Skip it anyway, for now.

Use Claude in your browser for the first week or two. Fewer moving parts, nothing installed, nothing to uninstall if you decide this is not for you. Once it has earned a permanent spot in your week, install the desktop app or the phone app then.

On the phone, when you get there: iPhone users open the App Store, Android users open the Play Store, and search for Claude. Check that the maker is listed as Anthropic. There are lookalike apps sitting right next to the real one wearing similar names and similar icons. Read the maker's name before you download, the same way you read the name on a badge before you believe the uniform.

Step 5: The privacy switch almost everybody misreads

Now the setting that matters most, and the one I see people misunderstand constantly.

You will see a toggle asking whether your conversations can be used to help improve the model.

Turn it off if you want. I did. It is the right instinct.

Just do not mistake what it means. People flip that switch off and conclude their information is now sealed. That is not what happened.

Here is the actual line. Anything you type into any AI system that is not running on your own machine leaves your building. It travels to a server owned by somebody else. It sits there. It can be reached with a subpoena. The training toggle governs one specific thing, whether your conversations get used to train future models. It does not govern where your words went.

The only genuinely private setup is a model running on hardware you own, and that is real hardware. NVIDIA sells a desktop AI machine called the DGX Spark in the 5,000 dollar range. AMD has a comparable box closer to 5,700. A loaded Mac Studio lands in similar territory. Those machines run a model locally, where nothing leaves the room.

For almost everybody reading this, that is not the setup. You are using a system that belongs to somebody else, and that is completely reasonable. Just operate accordingly.

So here is the working rule. Everyday questions, drafting, research, learning, planning, all of it, absolutely. Social Security numbers, bank and account numbers, passwords, executed contracts you are not free to share, and anything covered by medical, credit, or financial privacy law, none of it, ever. Not because this company is out to get you. Because that is basic street sense on any website.

Step 6: The introduction screen

Claude now asks your name and what you do. Writer, educator, engineer, health care, finance, legal, student, founder, consultant, scientist, and a good long list beyond those, including simply Other.

Pick something. This is not a commitment.

What that choice does is shape how Claude talks to you and what it suggests. Pick educator and it leans toward explaining concepts. Pick scientist and it leans toward experiments and analysis. Pick operations and it starts offering to draft procedures and vendor requests.

Here is the part worth knowing: you can override it any time by just saying so. Type "I made a mistake at setup, I am not an educator, I am a rocket scientist" and it adjusts on the spot. There is no settings menu to hunt through. You talk to it, and it follows.

That is the thing to internalize about this whole tool. It is not software with buttons you have to find. It is closer to working with a very sharp assistant who listens and adjusts.

Step 7: The first four things to type

You are in. There is a text box at the bottom. Here is where to start, in order.

One. Ask it how to work with it.

Hello. Can you tell me the best way to use you?

It will tell you. Be specific about context, iterate instead of starting over, bring it the whole workflow rather than one isolated question. Straight from the tool itself.

Two. Teach it who you are.

Can you help me teach you about how I like to do things?

This one matters more than it looks, especially if you will produce anything carrying your name. Tell it how you write, what you care about, what you never say. The gap between generic AI output and something that sounds like you is almost entirely this step, and almost nobody does it.

Three. Ask what the buttons mean.

You will see Projects, Artifacts, Code, Customize along the side. Do not go looking for a manual. Type the question:

What does Projects mean, and when would I use it?

It will walk you through its own interface. Ask about anything you see.

Four. Make it come down to your level.

This is the single most useful sentence I have handed anybody, and I use it myself constantly:

Explain that to me like I am in the 7th or 8th grade.

No offense in it at all. When you strip the jargon out, the light comes on dramatically faster. I ask for that on subjects I already know reasonably well, because the plain version is usually the version that actually sticks. Ask for it as many times as you want. It never sighs, never rushes you, never makes anybody feel slow.

That patience is not a side effect. It is the entire reason this thing is useful to people who were told for thirty years that they were bad with computers.

Where Claude sits among the others

Quick map, so the names stop being noise.

Anthropic builds Claude. OpenAI builds ChatGPT. Microsoft has Copilot. xAI has Grok. Meta, which is Facebook, has its own as well. There are more.

As of August 22, 2026, the two running out front for general use are Claude and ChatGPT.

One real difference worth knowing at signup. Claude states plainly that it does not show ads and does not let advertisers influence its answers. ChatGPT began showing ads to free and lower tier users in the United States in February 2026. OpenAI's position is that those ads are clearly labeled and do not change the answers themselves, and their paid tiers stay ad free. Reasonable people are watching how that plays out. It is simply worth knowing which model you are talking to and how it is funded.

If you want to go deeper on how Claude specifically was trained, look up constitutional AI. That is a genuine rabbit hole and a fascinating one. Not required reading to use the tool. Interesting if the wiring interests you.

The other thing to understand: these systems are getting sharper month over month. Some of the most capable versions are not released to the general public at all. They go to large organizations first, under tighter terms, while the public gets a more constrained version. That is a longer conversation for another day, but it explains why the ground keeps shifting under everyone.

Bonus: the link structure I actually use

One more thing, for anyone who wants to go a step further than the setup itself. I recorded a short screen walkthrough of the actual link structure I use to keep my own AI tools, bookmarks, and resources organized so nothing gets lost once you start using this daily. It is a companion to everything above, not a replacement for it.

What to actually do tonight

Do not try to master this. Do one loop.

Go to claude.ai. Make the free account. Skip the download. Turn off the training toggle. Pick any role. Then ask it one real question about something genuinely on your plate this week, and when the answer comes back too technical, tell it to explain like you are in the 8th grade.

That is the whole first session. Fifteen minutes.

The people the internet left behind were never the older ones. They were the ones who never sat down and tried. Sitting down is the entire skill.

If you want help going further with this, that is exactly what we do here. Text the word AI to (661) 476-2217, or book a time and we will walk your setup together.

Common questions

Connor T. MacIvor · CalDRE #01238257 · Sync Brokerage, Inc. · DRE #02031490