Candidates are told when they're interacting with AI: the screening chat carries a disclosure, a human reviews every conversation before any decision, and candidates can request a human or opt out of automated processing at any time. This isn't fine print; it's built into the product and can't be switched off.
| Channel | What the candidate is told |
|---|---|
| Web screening chat | The line under the assistant's name always says it is an AI ("AI Screening Assistant" or "Automated assistant (AI)"). The opening message says it is an automated assistant and that a human recruiter reads every conversation. About & privacy opens "You are chatting with an AI", which explains that answers may be assessed with AI and that a human recruiter reads every conversation before any decision. |
| AI-assisted emails from your mailbox | Sequence emails sent from your connected mailbox carry an AI-assistance notice in the footer. It cannot be disabled in settings. A note you type yourself to one candidate carries no such notice, because no AI drafted it. |
| Cold first-touch sourcing emails | No AI notice. These are fixed, human-written templates, not AI-generated messages. The candidate is told they are dealing with AI when they open their screening link. |
And a simple rule underneath all of it: the AI never claims to be human. If a candidate asks "am I talking to a bot?", it says yes.
From 2 August 2026, Article 50 of the EU AI Act requires that people be informed when they're interacting with an AI system. The disclosures above are built in and not optional, so there is nothing for your agency to switch on. The screening chat, where the candidate actually talks to the AI, discloses it up front. Cold first-touch emails carry no notice, as described above. Settings → Compliance shows this row as "Partial" for that reason; whether that is enough for your use is a question for your own counsel.
PlacementFlow screens candidates through text only — a web chat, reached by a private link. There are:
This is a deliberate design decision, not a missing feature. Voice-based AI screening carries serious legal exposure (the EU AI Act prohibits emotion-recognition in employment contexts, and voiceprint collection triggers biometric-privacy laws in several jurisdictions) and candidates broadly dislike it. Text keeps the interaction transparent, reviewable, and on the candidate's own terms.
This implements GDPR Article 22 — the right not to be subject to purely automated decision-making. Practically, PlacementFlow goes further: even without an opt-out, a human recruiter reviews screening outcomes before a candidate advances or is rejected.
From the candidate portal, a candidate can view their AI decision history and download a PDF report of decisions made about them — plus request erasure of their data. Transparency isn't just a disclosure at the door; it extends to the record afterwards.
You get to answer candidate and client questions about AI honestly and specifically: candidates are told, humans review, opt-outs work, and everything is logged. If a candidate raises an Article 22 or transparency concern, the machinery to honour it is already running — you don't have to build a process.