AI Interview Red Flags: Ethics, ChatGPT, and When to Walk Away from the Job
Two sets of red flags exist here. Yours (using AI dishonestly) and the employer’s (using AI illegally). Both are covered plainly.
Can You Use ChatGPT During an AI Interview?
The honest answer depends on what “use” means and which platform you are on.
Text chat (Sapia, Mya, Paradox)
Sapia explicitly detects AI-generated content. From their candidate documentation: they use linguistic fingerprinting and anti-plagiarism scanning to identify responses that were not written by the candidate. This is not theoretical - it is documented. Submitting a ChatGPT response to Sapia is likely to be flagged.
Even on platforms without explicit detection, AI-generated text tends to be generic and structured in ways that score lower on BARS rubrics, which reward specific examples and authentic language that generic AI output typically lacks.
One-way video (HireVue, Willo, etc.)
“Live-assist” tools (Final Round AI and similar) that listen to the audio of your session and feed you suggested answers exist and are used by some candidates. Detection is an ongoing arms race: some employers use second-camera proctoring, some monitor browser focus and keystroke patterns. HireVue has not publicly confirmed specific detection measures in 2026, but the proctoring layer is tightening.
More practically: the role you land on the back of a generated answer is a role you may not be able to do. Offer rescission post-hire-and-discovery is a real career consequence, separate from the detection risk during the interview itself.
The practical advice, plainly:
Prepare with AI. Use ChatGPT to generate practice questions, refine your STAR stories, improve your phrasing in the days before the interview. That is legitimate preparation, identical in kind to using an interview coach or a prep book. Do not submit AI-generated answers as your own. The detection risk is real on some platforms; the authenticity deficit is real on all of them; and the role-fit risk is real at every employer.
The Legal Grey Area
There is no US federal law criminalising a candidate’s use of AI during a job interview in 2026. Sapia’s anti-plagiarism detection is contractual - you agreed to their terms of service, and submitting generated content may constitute a misrepresentation. The consequences are contractual (disqualification, offer rescission) rather than legal (criminal liability or civil fine).
UK and EU law are similar: no specific statute criminalises candidate AI use in interviews. GDPR applies to how the employer handles your data; it does not restrict what you can type into a chat window. The risk remains reputational and contractual, not criminal.
How Sapia’s Anti-Plagiarism Detection Works
Based on Sapia’s publicly documented candidate guide (verified April 2026): Sapia uses a combination of text fingerprinting and behavioural signals to detect AI-generated responses. The fingerprinting compares response patterns against known AI-generation signatures. Behavioural signals include the speed and edit pattern of typing (consistent with pasting vs. composing). Cross-reference against generator-typical sentence structures and vocabulary patterns is also applied.
This is not infallible detection, but it is good enough that direct paste-and-submit of ChatGPT output will typically trigger flags. Lightly edited AI output may not always be detected, but loses the authentic specificity that makes responses score well regardless.
Second-Camera and Keystroke Detection
An emerging 2026 reality in high-stakes assessment contexts: some platforms (Talview most prominently) add remote proctoring that monitors browser focus, keystroke patterns, and can require a secondary camera showing the room environment. These are most common in academic-adjacent assessments and government/military hiring contexts.
For standard HireVue or Willo commercial hiring, secondary proctoring is less common but employers can activate it. If your invite mentions “proctoring” or “environment check” or “Talview”, treat those as active proctoring sessions.
When the Employer Is the One with the Red Flag
Some employer behaviours in the AI interview process are legally required, and their absence is a signal worth taking seriously. These are the red flags on the employer side.
No disclosure of AI use where legally required
If you are a NYC resident or applying for a NYC-based role and the employer uses an AI interview system without giving you at least 10 business days notice and explaining what characteristics it evaluates, they are non-compliant with Local Law 144. In Illinois, the AIVIA standalone notice requirement applies. Non-compliance suggests weak HR governance across the board.
No published bias audit summary from a NYC-hiring employer
NYC Local Law 144 requires that employers publish an independent annual bias audit summary on their website before using an AEDT. If you cannot find this on the employer's careers page or on request, they are potentially non-compliant. You can ask HR for the URL directly. A missing bias audit suggests either non-compliance or a very recent adoption that has not yet been audited.
No alternative selection process offered when one is required
In NYC, employers must offer instructions for requesting an alternative selection process or reasonable accommodation as part of the AEDT notice. If their notice exists but omits this, or if HR refuses to engage when you request one, that is a compliance failure and a broader signal about how the organisation handles process and legal obligations.
No accommodation mechanism when requested
ADA and Equality Act 2010 require good-faith engagement with reasonable accommodation requests in all jurisdictions, regardless of AI law. An employer who does not respond to, dismisses, or ignores your accommodation request (especially if specific and written) is signalling poor HR practice. Withdrawing your candidacy for requesting accommodation is unlawful.
A platform that claims to score things that are not appropriate or that they legally should not score
If an employer's documentation, or the platform's own materials, suggest that the AI is evaluating "personality type" from voice tone, "cultural fit" from facial micro-expressions, or similar claims, that is a red flag. Ask specifically what characteristics their AEDT evaluates. Under NYC and Illinois law, this information must be disclosed. Vague answers ('our AI assesses fit') combined with refusal to specify characteristics suggest either non-compliance or practices they know would not withstand scrutiny.
Opaque automated decision with no right to explanation
Under the EU AI Act from 2 August 2026, candidates in EU jurisdictions have the right to meaningful information about automated decisions affecting them. If an employer refuses to provide any explanation of how the AI scored your interview or what the scoring showed, that is a compliance issue in EU contexts. Outside the EU, it is a signal about the employer's transparency culture.
The advice, plainly:
Prepare with AI. Do not submit AI-generated answers. You will score better on your own honest STAR-structured examples than on anything ChatGPT generates, because BARS rewards specificity and authentic language that AI-generated text consistently lacks. And if the employer is showing the red flags above, that tells you something about the culture you would be joining.