For hiring managers

What is AI sourcing? How AI finds and contacts candidates who have not applied, and the rules that apply

On this page
  1. How does AI sourcing work?
  2. What kinds of AI sourcing tools are there?
  3. How is AI sourcing different from Boolean search, AI screening and rediscovery?
  4. What laws apply to AI sourcing?
  5. What should you ask before using an AI sourcing tool?
  6. What should staffing agencies know about AI sourcing?
  7. A copy-ready AI sourcing brief
  8. Questions people ask

AI sourcing is the use of artificial intelligence to find potential candidates who have not applied for a role, rank them against what the job needs and help contact them. A recruiter describes the role in plain language or pastes the job description; the tool turns that into search criteria, searches a profile network, the open web or the company's own applicant tracking system, explains why each person matches and drafts outreach for the recruiter to send. As of November 2026, LinkedIn says its AI-assisted search in Recruiter lets recruiters "copy and paste your intake notes or job description" and is optional, with Boolean search still available, and New York City's enforcement agency says its AI hiring law's bias audit and notice rules do not apply to tools used only to "conduct outreach to potential candidates."

This page defines the term and the rules around it. Agents that run the whole loop, from search to scheduling, and the products sold as agents are on what is agentic recruiting; the wider map of AI in hiring is on what is AI in recruiting.

How does AI sourcing work?

  1. Intake. The recruiter gives the tool the role: a job description, intake notes or a typed request such as "warehouse supervisors with WMS experience within 30 miles of Columbus" (an example).
  2. Query building. The tool turns the request into filters and search terms. LinkedIn says its advanced AI-assisted search will "analyze what a user wrote and instantly populate the right search filters" (LinkedIn AI explainability page, as of November 2026).
  3. Retrieval. It searches whatever data it has access to: a professional network, profiles compiled from public web sources, or past applicants in your own ATS or CRM.
  4. Ranking and explanation. It orders the results and, in better tools, says why. LinkedIn says its tool identifies which qualifications appear in a candidate's profile and which are missing.
  5. Outreach. It drafts a personalized message, and some tools send sequences on a schedule.
  6. Human review and feedback. The recruiter keeps or discards people, edits and sends messages, and the tool may learn from those actions. LinkedIn says its Hiring Assistant "does not make autonomous or automatic decisions."

What kinds of AI sourcing tools are there?

KindWhat the recruiter gives itWhat it returnsMain thing to check
Natural-language search inside a profile networkA job description or plain-language requestMatching profiles with the filters it chose, which the recruiter can editWhich filters it set, and whether you would have chosen them
Open-web profile aggregatorsSkills, titles, locationsProfiles compiled from public sources, often with contact detailsWhere the data came from, on what terms, and how a person opts out
Talent rediscovery in your ATS or CRMA new requisitionPast applicants and silver medalists who fitWhether old notes and outcomes bias who it surfaces; see silver medalist candidates
Outreach draftingA shortlist and the rolePersonalized messages and follow-upsAccuracy: a message that misstates a person's background or the job's pay is your message
Sourcing agentsA goal and limitsSearches, shortlists and messages, repeated over daysWhat it may do without approval; see agentic recruiting

How is AI sourcing different from Boolean search, AI screening and rediscovery?

TermWho is being looked atHow the search works
AI sourcingPeople who have not appliedThe tool interprets the request and can match related skills that are not written word for word; LinkedIn says its search can match "harder-to-define skills not explicitly listed on resumes"
Boolean sourcingPeople who have not appliedExact terms and operators the recruiter writes; see the role pages such as Boolean search strings for recruiters
AI screening or rankingApplicants for a specific jobScores or orders the people who applied; the step most AI hiring laws target
Talent rediscoveryPeople already in your databaseMatches past applicants to new roles, by AI or by search
Recruitment marketingEveryone who might see your jobsAds, content and campaigns rather than person-by-person search

The two kinds of search are not rivals. LinkedIn's own page says "you can still do Boolean searches in Recruiter" and that its AI can create or modify Boolean strings (LinkedIn, AI-Assisted Search and Projects, as of November 2026). A Boolean string is still the clearest record of exactly what you searched for, which matters if anyone later asks why certain people were never contacted. Who usually does this work, and how the sourcer role differs from the recruiter's, is on sourcer vs recruiter.

What laws apply to AI sourcing?

Not legal advice. This section summarizes the statutes, city and state texts, agency pages and terms linked below as published in November 2026. Which law applies depends on where the job and the people contacted are.

  • Title VII reaches sourcing. An employer may not "limit, segregate, or classify" applicants in a way that deprives anyone of opportunities because of race, color, religion, sex or national origin, and an employment agency may not "fail or refuse to refer" or "classify or refer" anyone on those bases (42 U.S.C. 2000e-2(a)(2) and (b)). Neutral filters that screen out a group can be challenged as disparate impact under 2000e-2(k); see disparate treatment vs disparate impact.
  • Age signals in outreach and filters. The age law bars notices or advertisements indicating "any preference, limitation, specification, or discrimination, based on age" (29 U.S.C. 623(e)). Graduation-year and "years since degree" filters are the usual way an AI search drifts into age; see age discrimination in hiring.
  • New York City. The Department of Consumer and Worker Protection answers "No" to whether Local Law 144 applies when a tool is used to "scan a resume bank, conduct outreach to potential candidates, or invite applications," because "the bias audit and notice requirements do not apply" to someone who "has not applied for a specific position" (DCWP FAQ). The rules can apply once those people apply and a tool scores them; see NYC Local Law 144.
  • Illinois covers recruitment. Since January 1, 2026 it is a civil rights violation, "with respect to recruitment, hiring, promotion" and other decisions, for an employer to use AI that has the effect of subjecting employees to discrimination, or "to use zip codes as a proxy for protected classes" (775 ILCS 5/2-102(L); detail on Illinois HB 3773). Other states are on AI hiring laws by state.
  • Data protection for people you did not collect data from. Under the UK GDPR, when you obtain personal data from another source you must give privacy information "within a reasonable period of obtaining the personal data and no later than one month," or at the first communication if you use it to contact the person (ICO, Right to be informed). The EU rule for sourced candidates is covered in the GDPR checklist for recruiters.
  • The FCRA, if a vendor sells profiles for employment decisions. A company that compiles information on people and furnishes it to employers for hiring decisions can be a consumer reporting agency; see what is a consumer reporting agency and what is the FCRA. The CFPB's 2024 circular on background dossiers and algorithmic scores for employment was among the guidance it withdrew in May 2025 (90 FR 20084); the statute itself did not change.
  • Platform terms. LinkedIn's User Agreement (effective November 3, 2025) forbids using "software, devices, scripts, robots or any other means or processes" to "scrape or copy the Services, including profiles and other data from the Services" (section 8.2). A tool that works by scraping puts the recruiter's own account at risk.

What should you ask before using an AI sourcing tool?

  1. Where does the profile data come from, under what terms, and how often is it refreshed?
  2. Can I see and edit the criteria the tool built from my request, and does it explain each match?
  3. Which fields can it filter or rank on? Ask specifically about graduation year, age-related fields, names, photos, addresses and zip codes, and switch off anything that stands in for a protected trait.
  4. Does it learn from our past hires or recruiter clicks, and can that be turned off or limited to one role? LinkedIn, for example, says its Hiring Assistant "learns from recruiter actions" such as which candidates are viewed, contacted or advanced, and that those learnings "are confined to the user and the particular role."
  5. Does anything go out without approval? Messages, connection requests and follow-up sequences should need a person's click unless you have decided otherwise in writing.
  6. How are opt-outs handled, and does a "do not contact" flag follow the person across every sequence and user?
  7. What is kept, where, and for how long, including messages and the tool's match explanations?
  8. Has the vendor tested who its results surface across groups, and will it share the method?

What should staffing agencies know about AI sourcing?

Agencies source for many clients at once, so the same tool runs under many different instructions. Title VII's employment agency clause applies to how an agency classifies and refers people, and a client's request for "recent graduates only" or "no one from that zip code" does not become acceptable because a tool executes it. Keep the search brief for each client in writing, use the job's real requirements as criteria, and record who approved changes. Data terms matter more for agencies too: if the agency adds sourced profiles to its own database, it holds that personal data for every future search, not just the one client's. Outreach wording for passive people is on how to recruit passive candidates, and for past applicants on candidate rediscovery email templates.

A copy-ready AI sourcing brief

AI SOURCING BRIEF (example; one per requisition)
Role: ______________  Req #: ______  Client (agencies): ____________
Must-haves (job-related, from the job description):
  1. ____________  2. ____________  3. ____________
Nice-to-haves: _____________________________________________
Location rule: [ ] onsite ____ miles of ______  [ ] remote in: ______
Filters NOT to use: graduation year, age, names, photos, zip code
  as a stand-in for anything other than commute: ________________
Data sources the tool may search: [ ] network [ ] our ATS/CRM [ ] web
Outreach: [ ] recruiter approves each message  [ ] approved template only
Opt-out / do-not-contact list checked?  [ ] yes
Criteria the tool generated (paste them): ________________________
Reviewed by: ________  Date: __/__/____  Changes approved by: _____

The brief is an example to adapt. Pasting the tool's generated criteria into the brief gives you the same kind of record a Boolean string does, and comparing the two is a quick way to see when the tool has read more into the request than you meant.

Questions people ask

What is AI sourcing in recruiting?

AI sourcing is the use of artificial intelligence to find potential candidates who have not applied for a role, rank them against what the job needs and help contact them. Typical tools let a recruiter describe the role in plain language or paste a job description, search profile databases, the open web or the company's own ATS, explain why each person matches and draft outreach messages for the recruiter to review.

How is AI sourcing different from Boolean search?

Boolean search finds profiles that contain the exact words and operators a recruiter types, so the recruiter controls every term. AI sourcing interprets a plain-language request or job description, turns it into search criteria and can match related skills that are not written on a profile word for word. LinkedIn, for example, says its AI-assisted search is optional and that Boolean searches still work in Recruiter.

Does NYC Local Law 144 apply to AI sourcing tools?

Usually not to the sourcing step itself. New York City's enforcement agency says the bias audit and notice requirements do not apply when an employer or agency uses a tool to scan a resume bank, conduct outreach to potential candidates or invite applications, because the law covers people who have applied for a specific position. Once those people apply and a tool scores or ranks them, the law can apply.

Is it legal to scrape LinkedIn profiles for recruiting?

LinkedIn's User Agreement forbids using software, scripts, robots or other means to scrape or copy the service, including profiles. Breaking it risks the recruiter's account, whatever the legal position on scraping in general. Ask any AI sourcing vendor where its profile data comes from and on what terms, and get the answer in writing.

Can AI sourcing tools be biased?

Yes. A tool that learns from past hires or filters on proxies such as graduation year or zip code can steer outreach away from protected groups, and Title VII covers how employers and employment agencies classify and refer candidates. Keep search criteria job-related, remove proxy filters, review who the tool surfaces, and record why people were or were not contacted.