Is resume screening by AI legal in the US?

It depends on the state and city. NYC has specific requirements (AEDT Law) for automated employment decisions including bias audits and candidate notification. California AB-2930 expands similar requirements. We scope screening agents narrowly and pair them with the required notices and audits.

What about the EU AI Act?

The EU AI Act classifies employment-related AI as 'high-risk.' If you have EU candidates, you need transparency, human oversight, and record-keeping. Workspace Agents support the record-keeping side; you still need the governance.

Can the agent handle sensitive conversations like terminations?

No. Hard rule. Separations, PIPs, comp discussions, harassment reports — all human-only. The agent's value is saving HR time on routine work so they have more time for the hard conversations.

Will this help or hurt my employer brand?

Depends on transparency. Candidates who discover a bot screened them without disclosure react badly. Candidates who see a well-designed agent answering their FAQ at 10pm Friday are impressed. Tell people what the agent does; over-communicate.

OpenAI Workspace Agents for HR Teams

HR is simultaneously the most tempting and most fraught function to automate. Tempting because so much of the work is repetitive policy-and-paperwork. Fraught because HR decisions are high-liability and AI bias is a real risk. Workspace Agents can take meaningful load off HR teams if you're disciplined about what you automate versus what stays human. This page walks through the safe starting points and the hard lines.

Where agents earn their keep on a hr team

Internal policy and benefits Q&A

Every HR team fields the same 20 questions constantly — PTO policy, benefits enrollment, expense rules, remote work guidelines. An agent grounded in your policy docs answers these instantly in Slack. HR handles the edge cases.

Resume-to-rubric screening (first pass only)

Candidates upload resumes to an ATS. The agent scores them against a job-specific rubric (required skills, years experience, role match) and flags the top tier. A recruiter reviews every flagged resume. The agent never auto-rejects.

Onboarding concierge

New hire day one: the agent answers questions about the tool stack, points them to the right docs, schedules intros to key people. Saves HR 3–4 hours per new hire and improves the onboarding experience.

Interview kit prep

Before every interview loop, the agent assembles the kit — job spec summary, candidate background, suggested questions per interviewer based on competencies to probe. Hiring managers show up prepared.

Rollout playbook

01
Start with internal-only, read-only agents
Policy Q&A for employees. No external communication. No decisions. This is the least-risky agent in HR — get comfortable here first.
02
Do a bias audit before screening
Before any agent touches candidate data, audit for bias. Run the agent on 100 historical resumes where you know the outcome. Compare demographics in the flagged set to the unflagged set. Document the audit. Redo annually.
03
Hard rules for candidate-facing work
The agent never sends rejection emails, never decides finalists, never makes comp recommendations. It drafts, ranks, surfaces — humans decide. Encode this in the system prompt and in the connector permissions.
04
Annual compliance review
HR regulations change (NYC AEDT, California AB-2930, EU AI Act). Review agent workflows yearly for new compliance obligations. Update the audit, the bias test, and the disclosure to candidates.

Agents I build for hr teams

Questions

Plan a hr-team rollout

20-min intro call. I'll sanity-check your stack and give you a sequencing recommendation.

Book a 20-min intro call Or send a note

For other teams

free tools

Turn this into something actionable

Read the full OpenAI Workspace Agents setup guide →