AI That Forgets: Why We Built Ask Safely

AI is actually pretty useful

Let's start with something that might surprise you coming from a company focused on AI safety: we think AI is genuinely helpful.

It can explain algebra to your eighth-grader at 9pm when you've forgotten everything about quadratic equations. It can help you draft that difficult email to your boss. It can answer the parenting question you're too embarrassed to ask anyone else, or help you figure out what that weird rash might be before deciding whether to call the doctor.

The productivity benefits are real. And if you're like most families we talk to, someone in your household is probably already using ChatGPT, even if they feel a little uneasy about it.

That unease is what we want to talk about.

The problem nobody's talking about

Here's what most people don't realize: ChatGPT remembers everything. Every question you ask, every worry you share, every vulnerable moment you type into that box is stored, analyzed, and used to build a profile of who you are.

One of our early users put it perfectly: "I literally uploaded my wife and my draft estate plan to ChatGPT and regretted it."

Think about what you've asked AI in the past month. Maybe it was a health concern you hadn't told anyone about yet. A question about your kid's behavior that worried you. Financial stress you were trying to work through. A relationship issue you needed to think out loud about.

These aren't just queries. They're windows into your psychology: your fears, your insecurities, your family's private struggles. And they're being collected to build what we call a "digital twin": a psychological profile that belongs to the AI company, not to you.

Why this matters—now and later

Here's the simplest way we can explain it:

Imagine keeping a private journal. You write your real thought…the stuff you'd never say out loud. Now imagine that every night, someone from the company that sold you the notebook lets themselves into your home, photographs every page, and takes the images back to a team that highlights your fears, your insecurities, and your family's private struggles. Then they sell those highlights to advertisers who want to know exactly which buttons to push.

That's essentially what's happening or will happen when you use most AI tools today. It's what keeps us up at night: this isn't just about today's risks. It's about what becomes possible tomorrow.

What happens when AI systems know your teenager's deepest insecurities? When they've mapped your family's anxieties, your unprocessed stress, the questions you ask at 2am when you can't sleep? That information doesn't just sit in a database. It becomes a tool—for advertising, for manipulation, for purposes we can't even predict yet.

We're not being paranoid. We're pattern-matching. And the pattern says: once this data exists, it gets used in ways nobody anticipated.

How Ask Safely does it differently

We built Ask Safely on a simple principle: your AI assistant should help you, on your terms. Not study you. Not hide anything from you. Not try to befriend and manipulate you. Not let us lean into delusions in moments of weakness.

Here's what that means in practice:

Amnesia by design. Your conversations automatically delete after 8 hours. Not archived. Not anonymized. Deleted. If you want to save something, you can extend the self-deleting window—but the default is forgetting, not remembering.

One user told us: "I started to ask ChatGPT about some mental health questions but closed the app and switched to Ask Safely instead." That's the feeling we're going for. The freedom to ask anything without wondering who's watching.

Your data never trains our AI. Ever. This isn't a setting you have to find and toggle. It's how we're built.

We chose Anthropic for a reason. Ask Safely runs on Claude, made by Anthropic. We're proud of that choice. Anthropic is the only major AI lab that puts safety research at the center of everything they do. Their mission aligns with ours: AI should benefit people, not exploit them. When you use Ask Safely, your conversations are processed securely and never used to train models—ours or anyone else's.

Simplicity is a feature. No model selection. No confusing privacy settings buried in menus. No choices you need a PhD to understand. We handle the complexity so you don't have to.

As one of our users in Arizona told us: "I'm a 9 out of 10 AI skeptic, but this just works. I know my data is deleted and I don't have to deal with Google links anymore."

Who we are and why this matters to us

I'm Mike, and I'm a dad of a two year old, moonlighting as the Co-Founder and CEO of Ask Safely based on Minneapolis Minnesota.

Before starting Ask Safely, I spent years at Walmart leading the team that built their grocery delivery business from zero to tens of millions of families. I learned a lot about scale, about what it takes to serve busy families on a budget, and about how companies think about consumer data.

That last part is what led me here.

I've spent over a decade thinking about data business models. How companies collect information, what they do with it, and what it means for the people on the other end. And when I started using AI tools hourly in my life, I felt that same unease you probably feel. The benefits were obvious. But so was the future cost I was asking my future self and my family to pay.

I couldn't find an AI tool I actually trusted with my family's questions. With my most sensitive information. So we built one.

My co-founder Cass comes from Target, where she led UX for AR/VR experiences. Her whole career has been about one thing: designing technology that respects people. She says it better than I can: "People ignore design that ignores people."

We built Ask Safely because we believe families deserve AI that works for them not AI that works on them.

Try it and see how it feels

We're not asking you to trust us because of what we say. We built Ask Safely so you don't have to take our word for it.

Use it. Ask the questions you've been hesitant to ask elsewhere. Notice how it feels to use AI without that weight in the back of your mind.

Your chat. Your data. Your control.

Have questions about how we protect your privacy? Want to know more about our approach? We're always happy to talk—reach out at mike@asksafely.ai.