LIVE
Crustdata now works inside Claude
Give Claude real-time people and company data with MCP.
LIVE
Crustdata now works inside Claude
Give Claude real-time people and company data with MCP.
Crustdata now works inside Claude
CANDIDATE FRAUD DETECTION
CANDIDATE FRAUD DETECTION
Spot fake candidates before you get on the call
Spot fake candidates before you get on the call
Spot fake candidates before you get on the call
Fake and AI-generated applicants are everywhere now, and most of them look fine on paper. Crustdata gives you the signals to catch them early, like when a professional profile was actually created and whether the person shows up anywhere else online. You see the flags and the reasons behind them. You decide who's real.
Fake and AI-generated applicants are everywhere now, and most of them look fine on paper. Crustdata gives you the signals to catch them early, like when a professional profile was actually created and whether the person shows up anywhere else online. You see the flags and the reasons behind them. You decide who's real.
YOU
YOU
Is this applicant real? Show me why.
Is this applicant real? Show me why.
Jordan Mercer
Jordan Mercer
Senior Backend Engineer
Senior Backend Engineer
Applied · remote role
Applied · remote role
Likely fake
WHY IT'S FLAGGED
Profile created 3 months ago
Claims 12 years of experience
0 verifications
4 connections
Priya Sundaram
8y history • GitHub and Scholar match
8y history • GitHub and Scholar match
✓ Likely real
Hiring has a fraud problem, and it keeps getting worse
Hiring has a fraud problem, and it keeps getting worse
Hiring has a fraud problem, and it keeps getting worse
AI made it easy to fake a resume, a profile, even a live interview. Remote roles get flooded with people who aren't who they say they are. The recruiters we talk to are losing hours a day sorting real applicants from fake ones, and the fakes keep getting better.
AI made it easy to fake a resume, a profile, even a live interview. Remote roles get flooded with people who aren't who they say they are. The recruiters we talk to are losing hours a day sorting real applicants from fake ones, and the fakes keep getting better.
REAL DATA FROM PEOPLE WE SPOKE TO
REAL DATA FROM PEOPLE WE SPOKE TO
90-95%
90-95%
of applicants aren't real, on some roles
~30%
~30%
of applications are obvious scams, one recruiter told us
~$120k
~$120k
lost in a year to fakes who passed five or six interview rounds
2 hrs
2 hrs
a day, gone to manually checking profiles
The worst version of this is a fake getting caught at the reference stage, after your client already spent five interviews on them.
The worst version of this is a fake getting caught at the reference stage, after your client already spent five interviews on them.
A good fake looks just like a real person
A good fake looks just like a real person
A good fake looks just like a real person
You can't tell from the words on a page anymore. A well-built fake profile reads exactly like a real one. Three things make it hard:
You can't tell from the words on a page anymore. A well-built fake profile reads exactly like a real one. Three things make it hard:
The checks are manual. Recruiters go on gut feel, opening profiles, checking GitHub, and searching names one at a time.
The checks are manual. Recruiters go on gut feel, opening profiles, checking GitHub, and searching names one at a time.
The fakes keep improving. AI-written bios, stolen identities, made up jobs at companies that shut down.
The fakes keep improving. AI-written bios, stolen identities, made up jobs at companies that shut down.
It gets caught late. Background checks happen at the end, once the time and money are already spent.
It gets caught late. Background checks happen at the end, once the time and money are already spent.
The one thing that actually gives a fake away is cross-checking against real-world data, which is what we do.
The one thing that actually gives a fake away is cross-checking against real-world data, which is what we do.
YOU
YOU
Why does this profile look off?
Why does this profile look off?
WHY IT'S FLAGGED
Profile created 6 months ago
Claims 12 years of experience
Under 10 connections, no activity
No GitHub or Scholar footprint
First seen in the last 12 months
We give you the signals. You make the call.
We give you the signals. You make the call.
We give you the signals. You make the call.
Crustdata pulls the red flags from hundreds of data points and ranks the profiles that look off. For each one, you see exactly why it got flagged. No candidate data is stored on our side, and because you make the final call, your decisions stay yours to defend.
Crustdata pulls the red flags from hundreds of data points and ranks the profiles that look off. For each one, you see exactly why it got flagged. No candidate data is stored on our side, and because you make the final call, your decisions stay yours to defend.
WHY IT'S FLAGGED
Applicant 12
Likely fake
Why flagged
High risk
New profile, no network
Claims a decade of work
The red flags we surface
The red flags we surface
The red flags we surface
Profile creation date
Twelve years of experience on a profile made six months ago. That's usually the first tell.
Profile creation date
Twelve years of experience on a profile made six months ago. That's usually the first tell.
Connections and network
Real people build a network over years. Fresh fakes don't have one.
Connections and network
Real people build a network over years. Fresh fakes don't have one.
Profile verifications
Are their ID and recent employers verified, or is there nothing there?
Profile verifications
Are their ID and recent employers verified, or is there nothing there?
Activity history
Do they actually post and comment, or is the profile an empty shell?
Activity history
Do they actually post and comment, or is the profile an empty shell?
GitHub and Google Scholar
If they claim deep expertise, does it show up where it should?
GitHub and Google Scholar
If they claim deep expertise, does it show up where it should?
Email and identity patterns
Disposable domains, mismatched emails, the same person applying under different names.
Email and identity patterns
Disposable domains, mismatched emails, the same person applying under different names.
Have we seen them before?
Someone who only appeared in the last twelve months is worth a second look.
Have we seen them before?
Someone who only appeared in the last twelve months is worth a second look.
Profile detail
Generic, copied, or oddly thin profiles that read like a template.
Profile detail
Generic, copied, or oddly thin profiles that read like a template.
And we keep adding new signals as the fake profiles change tactics.
And we keep adding new signals as the fake profiles change tactics.
Two ways to use it
Two ways to use it
Two ways to use it
OPTION A
Inside Claude
Inside Claude
Inside Claude
For recruiters who want the checking done for them. We set you up with a Crustdata skill in Claude. Give it your candidates, from your ATS or a spreadsheet, and tell it what fake looks like to you, plus a few example fakes to learn from. It checks each candidate against our data and returns a ranked list of likely real or likely fake, with the reason for each.
For recruiters who want the checking done for them. We set you up with a Crustdata skill in Claude. Give it your candidates, from your ATS or a spreadsheet, and tell it what fake looks like to you, plus a few example fakes to learn from. It checks each candidate against our data and returns a ranked list of likely real or likely fake, with the reason for each.
YOU
Screen these 40 applicants
Screen these 40 applicants
Likely fake
✓ Likely real
Reason shown
OPTION B
Through the API
Through the API
Through the API
For teams building fraud checks into their own product or workflow. Pull the same signals through our API and set your own rules for what counts as fake. This is how ATS platforms and technical recruiting teams use it.
For teams building fraud checks into their own product or workflow. Pull the same signals through our API and set your own rules for what counts as fake. This is how ATS platforms and technical recruiting teams use it.
POST
/screen
200 · risk scored
GET
/signals/{profile_id}
200 · flags
Either way, check a whole list at once or check live as people apply.
Either way, check a whole list at once or check live as people apply.
Who uses this
Who uses this
Who uses this
Staffing and recruiting firms
Staffing and recruiting firms
In-house talent teams at AI labs, defense, and other high-security roles
In-house talent teams at AI labs, defense, and other high-security roles
ATS and recruiting platforms that want to hand their own customers fraud signals
ATS and recruiting platforms that want to hand their own customers fraud signals
VC and accelerator teams screening applicants
VC and accelerator teams screening applicants
Stop wasting your time on fake candidates
Stop wasting your time on fake candidates
Stop wasting your time on fake candidates
See how Crustdata flags fake candidates before they reach your team.
See how Crustdata flags fake candidates before they reach your team.
FAQ
Questions we hear
Questions we hear
Questions we hear
Do you verify government ID?
How is a profile creation date a fraud signal?
Can I use this with Greenhouse, Ashby, or my ATS?
Will it wrongly reject real people?
Does my candidate data stay private?
How is this different from a background-check company?
Do you verify government ID?
How is a profile creation date a fraud signal?
Can I use this with Greenhouse, Ashby, or my ATS?
Will it wrongly reject real people?
Does my candidate data stay private?
How is this different from a background-check company?
Products
Popular Use Cases
Competitor Comparisons
Use Cases
95 Third Street, 2nd Floor, San Francisco,
California 94103, United States of America
© 2026 Crustdata Inc.
Products
Popular Use Cases
Competitor Comparisons
Use Cases
95 Third Street, 2nd Floor, San Francisco,
California 94103, United States of America
© 2026 Crustdata Inc.
Products
Popular Use Cases
Competitor Comparisons
Use Cases
95 Third Street, 2nd Floor, San Francisco,
California 94103, United States of America
© 2025 CrustData Inc.