01
Known-bad recognition
A curated list of sites that exist to serve adult material, run a casino, or distribute malware. Small, hand-reviewed, and applied instantly.
Web safety
A good answer needsa clean library.
askFinz reads the web itself, which means deciding what deserves a place in it. Malware, scams, adult material and gambling are kept out. The harder job — the one this page is about — is doing that without quietly throwing away the good web along with the bad.
The short version
- Four layers decide, not one.
- We judge what a site is for, not what it mentions.
- Removing something needs two judgements to agree.
- Universities, hospitals and governments are never auto-removed.
- A person can overturn any of it.
How it works
04 layers · agreement requiredNo single check is trusted on its own, because no single check is reliable on its own. Each layer catches what the others miss, and the last one is a person.
02
Millions of reference domains
Industry threat and category lists covering malware, phishing, scams, fraud, adult, gambling and drugs. Refreshed every day, because new bad sites appear hourly.
03
A judgement call, twice
For the sites no list has ever heard of, we ask an AI what the site is for — then ask a second, deliberately opposite question. Both have to agree before anything is removed.
04
A person, with the final word
Every automated decision can be reversed by a human, and that override outranks everything above it. Nothing here is a black box that can't be argued with.
What we go by
the hard partFiltering is easy. Filtering fairly is not.
Any keyword blocklist can keep out pornography. It will also keep out the sexual-health charity, the addiction-recovery group and the medical reference — and nobody will notice until somebody needed one of them.
So the question we ask isn't what a site mentions. It's what the site is for.
Purpose, not subject
A site about a difficult subject is not the same as a site selling it. A sexual-health charity, an addiction-recovery group and a medical drug reference are all legitimate — and all of them would be caught by a filter that only looked at keywords.
Erring toward keeping
It's better to review a borderline site than to quietly delete a good one. Nobody notices a missing page until they need it, so the burden of proof sits with removal rather than inclusion.
Clean by default
Ordinary searches never surface adult or gambling material. It isn't a setting you have to find — it's how search behaves unless you deliberately change it.
Checked continuously
Sites change hands and get compromised. Reference lists refresh daily and pages are re-read over time, so a judgement made last year doesn't stand unexamined.
Live · the numbers
from the running systemSafety claims are easy to make and hard to check, so these come straight from the system rather than from a slide. “Hidden by default” means indexed but kept out of ordinary results — out of the way rather than erased.
Connecting…
—
Reference domains
checked against, refreshed daily
—
Kept out
refused before indexing
—
Hidden by default
indexed, out of normal results
—
Human overrides
wrong calls corrected
Questions
05- What exactly gets kept out?
- Sites whose purpose is to serve the thing: pornography, casinos and bookmakers, illegal drug sales, and malware, phishing or scam operations. Sites that discuss, treat, report on or help people with those subjects are not removed.
- Doesn't filtering just delete things you disagree with?
- That's the risk, and it's why the burden of proof sits with removal. Automated checks flag; they don't decide alone. A removal needs two independent judgements to agree, public-interest domains like universities and health services are exempt from automated removal entirely, and a person can reverse any of it.
- What if a legitimate site gets caught?
- It happens, and we treat it as a fault rather than an acceptable cost. Industry threat lists are aggressive by design — we've seen them flag a national bank, a university and a medical reference. That's why those lists inform a review rather than triggering a block, and why the human override exists.
- Can I turn the filtering off?
- Ordinary results are filtered by default. Material that's been indexed but hidden can still be reached deliberately — it's out of the way, not erased. Genuinely harmful content isn't indexed at all, so there's nothing to turn back on.
- How do I report something you've got wrong?
- Either direction — something that shouldn't be there, or something missing that should be — get in touch. Both are worth knowing about, and both get looked at by a person.
Seen something wrong?
Both kinds of mistake matter.
Something in the index that shouldn't be, or something missing that should be there — we want to hear about either. A filter nobody can correct isn't a safety feature, it's just an opinion nobody voted for.