# Bots and AI out of your survey, before you pay for them.

- Works in SampleLabs, Google Forms, Qualtrics or any other survey builder
- Tested with major panel providers, filtered out 20-40% of data as AI or inattentive answers
- See exactly why responses are flagged

Every response is checked before it joins your sample. You see what was removed, and why.

SampleLabs Protect checks every response to your survey for bots, scripts driving a real browser, AI-written answers, repeat visits and rushing, before the response counts. It guards surveys built in SampleLabs, and surveys in Google Forms, Qualtrics or any other tool through a short check page. Every check comes out of our own extensive research: we run our own bots against it and test it on real people before it can flag anyone. You choose whether flagged people are only marked or turned away, and you see the reason for every flag.

## What Protect checks

- **Bots that fill in forms.** A hidden field that people never see but bots fill in, plus Cloudflare's invisible check, the kind you pass on many websites without noticing.
- **Scripts in a real browser.** AI agents and bot tools that drive a real browser give themselves away in how they click, move the mouse and type.
- **Rented servers.** AI agents and bot tools mostly run on cloud servers. People answer from home, from work or on their phone.
- **A tampered bot check.** The browser's own note on how the survey was answered, checked against our clock and the answers that arrived.
- **AI-written answers.** Open answers that were pasted in, or read like they came from an AI tool, are flagged with the answer that gave them away.
- **Duplicates.** Repeat entries from the same browser or IP address are caught on the second try.
- **Attention check.** One question with a single right answer, like "Pick Disagree for this one." Someone who isn't reading gets it wrong.
- **Rushed answers.** Responses finished faster than anyone could read the questions are marked Rushed, so you decide whether to keep them.
- **Session recordings.** Watch any response back and judge for yourself. Typed text is hidden, and recordings are deleted after 30 days.

## How Protect works

1. **Build your survey, or bring your own.** Build it in SampleLabs, and Protect is a tab in the builder. Built it in Google Forms or Qualtrics? Paste its link into Protect and share Protect's link instead.
2. **Choose what happens to a flagged person.** Flag only lets everyone finish. Turn them away sends flagged people to an end page instead.
3. **See what was caught, and why.** Each flagged response shows its reason in your results, like "Clicked by a script". Turned-away responses stay out of your counts, charts and download.

## Protect pricing

Surveys built in SampleLabs are protected on every Build plan, Free included. For a survey built in Google Forms, Qualtrics or any other tool, you pay for each person Protect checks, billed monthly by card before VAT. Your first 10 checks a month are free, with no card needed. Each check costs less as your month grows:

- **First 20,000 checks a month.** $0.05 each.
- **20,001 to 50,000 checks a month.** $0.04 each.
- **50,001 to 150,000 checks a month.** $0.02 each.
- **150,001 to 300,000 checks a month.** $0.015 each.
- **Above 300,000 checks a month.** $0.01 each.

## Questions people ask

### How does SampleLabs Protect spot bots in survey responses?

Protect runs an ever-growing checklist of hundreds of checks, a mix of our own and trusted third-party ones, and we tune them every day against simulations of different kinds of bots. Some look at how the mouse and keyboard were used: scripts click the exact middle of buttons, slide the pointer in dead straight lines and type to a steady beat. People don't. Others look at traps only bots fall into, where a response comes from, and a range of other things. The main ones are explained on How we check for AI. https://samplelabs.io/how-we-check-for-ai

### Can Protect catch AI agents that use a real browser?

Yes. AI agents and tools like Browserbase drive a real browser, so they pass most "are you human" tests. Protect catches them by how the pointer and keys behave, by a bot check that's been tampered with, and by answers sent from a rented cloud server.

### Will Protect flag real people by mistake?

We do extensive research to keep that from happening. Every check is tested on real people before it can flag anyone, and a check that catches people too only takes notes.

### Does Protect judge people who don't use a mouse?

No. People who tap on a phone or tablet, use only a keyboard, or use a screen reader leave no mouse clicks to judge, so Protect's mouse checks never catch them.

### What happens to a response Protect flags?

You choose for each survey. With Flag only, everyone finishes and each flagged response shows its reason in your results. With Turn them away, flagged people are sent to an end page, and their response stays out of your counts, charts and CSV.

### Does Protect catch people rushing through a survey?

Yes. Protect marks responses finished faster than anyone could read the questions as Rushed, and once enough people have answered, responses far faster than the survey's usual pace too. Rushed is a quality label, not a bot flag: a quick person is still a person, so you decide whether to keep their answers.

### Is Protect a CAPTCHA?

No. A CAPTCHA asks people to prove they're human, and AI agents now pass most of them. Protect runs Cloudflare's invisible check too, but most of its checks ask nothing of people: they look at how the survey was answered.

### Can Protect check a survey in Google Forms or Qualtrics?

Yes. Paste your survey's link into Protect and share the link it gives you. People go through a short check page first, which runs the bot trap, rushed answers, duplicates and the attention check, then go straight on to your survey.

### Does Protect catch AI-written open answers?

Yes. On open-text questions, Protect flags answers that were pasted in or read like they came from an AI tool, and shows you which answer gave it away. Dot reads the answers itself, so this check runs on surveys built in SampleLabs.

### Can one person answer my survey twice?

Protect flags a second response from the same browser or IP address. With Turn them away, the second one never reaches your counts, charts or CSV.

### What does Protect record about the people answering?

Protect notes timings and positions on the survey page, never what anyone types. Session recordings hide typed text before it leaves the browser and are deleted after 30 days. The network address is kept rough, with its last part dropped, and the full address is only kept scrambled.

### How do I start with Protect?

If you already have a survey, paste its link into Protect and share the link it gives you. If not, build one in SampleLabs: Protect is a tab in the builder. https://samplelabs.io/protections/new