AI and automated decisions

Where automated systems are used

Singkatnya: Three places: scoring interviews, matching you to work and flagging content for a human to look at.

Interview assessment. Where you take a recorded or written interview, an automated system produces a score against a fixed rubric and a written rationale for each criterion.

Work matching. Which tasks appear in your queue is decided by your qualifications, your languages, your region and your availability.

Quality and content flagging. Automated checks compare submitted work against gold-set items and flag content in community spaces for review.

That is the complete list. If we add another, this page changes before it goes live.

What no automated system here does

Singkatnya: It never accepts you, rejects you, or reads your tone. One narrow exception on suspension and on face comparison, described below, rather than left for you to discover.

No automated system accepts an application, rejects an application, closes an account or withholds a payment. Every one of those is a decision made by a member of staff with a reason recorded against it. The one exception: an account strongly linked to one already suspended or banned (the same confirmed phone, identity document or payout account) is suspended automatically, because that is specifically the case of someone returning after a ban, and it can be appealed to a person who did not make the automatic decision.

There is no score threshold above or below which anything happens automatically, outside that one exception. There is no other queue an automated decision can enter without a person in it.

We do not perform expression analysis, emotion detection or voiceprint identification. We do perform one specific, narrow use of facial comparison: checking that the face in your identity-verification liveness clip matches the face on your identity document, and, where enabled, checking whether that same face appears on another account, solely to detect the same person operating more than one account — see the privacy notice's "What we collect" and "What we deliberately do not collect" sections for the full description. Neither comparison scores, ranks or judges you, and a recorded job interview's video is retained only so a person can confirm identity if something looks wrong; it is not processed by any automated system, and only a transcript of what you said is assessed.

We do not infer or record personality traits, character judgements, health, political opinion, religion, sexual orientation, trade union membership or activity or the emotional state of any contributor. Nothing of that kind is collected, derived or stored.

How interview scoring actually works

Singkatnya: Your audio is transcribed, the words are scored against four published criteria, and a person reads both.

The rubric is fixed before any interview is taken and is the same for every candidate for the same role. It has four criteria: accuracy and care, following instruction, judgement under ambiguity and clarity. Clarity assesses whether a reviewer could act on your answer, not eloquence and non-native phrasing is never penalised.

Your recorded answers are transcribed to text by a speech to text service. The transcript, not the audio and not the video, is what the scoring system reads.

The system returns a score and a written rationale per criterion. Both go to a member of staff along with your answers, and the staff member decides.

Where the transcription is low confidence, the reviewer is told to listen to the recording rather than rely on the score, because a low quality recording produces a low score that says nothing about the candidate.

Your rights over these systems

Singkatnya: Ask for a human review, ask for an explanation, ask us to delete your interview recording.

You may ask for a decision that involved an automated score to be reviewed by a different member of staff. Write to info@corpshore.ai and say which decision. We do not charge for this and it does not count against you.

You may ask for an explanation of the basis of any such decision, including the criteria applied and how your answers were assessed against them.

You may ask us to destroy your interview recording, and all copies held by us or by anyone acting for us, and we will do so within 30 days of the request. This right exists under the Illinois Artificial Intelligence Video Interview Act and we extend it to every applicant regardless of where they live, because there is no good reason to give an Illinois applicant a right we withhold from a Ghanaian one.

You may decline the recorded format. Tell us before you begin and we will provide a written interview instead, assessed against the same rubric. Declining the recording does not disadvantage your application.

You may withdraw consent to automated assessment entirely. In that case a member of staff assesses your written answers directly. This takes longer and we will tell you how long.

If you are applying from New York City

Singkatnya: We do not currently run the bias audit NYC law requires, so we do not use automated scoring on NYC candidates.

New York City Local Law 144 requires that an automated employment decision tool used on a candidate in the city has passed an independent bias audit within the previous year, that a summary of that audit is published, that candidates receive ten business days notice and that an alternative process is offered.

We have not yet commissioned that audit. Rather than use the tool and hope, applications from candidates in New York City are assessed by a member of staff reading the answers directly, with no automated score produced or used.

When the audit is complete we will publish its summary here and update this section. Until then, this is a statement of what we do not do rather than a claim of compliance.

If you work with us in the European Union

Singkatnya: The EU rules on algorithmic management describe how we already work. Here it is in their terms.

The EU Platform Work Directive (2024/2831) sets requirements for platforms that use automated systems to allocate, monitor or evaluate work. Member states must transpose it by 2 December 2026. The following describes our position in the Directive's own terms.

Restrictions on processing. We do not process data on your emotional or psychological state, your private conversations, your health beyond what you tell us voluntarily, your racial or ethnic origin, migration status, political opinions, religious beliefs or your trade union membership or activity. We do not attempt to predict any of these.

Human oversight. Staff monitor the effect of automated systems on working conditions, and the systems are reviewed rather than left to run.

Human review of significant decisions. Any decision that restricts, suspends or ends your access to work, affects your payment or affects your contractual status is made by a person. You may ask for an explanation of it and ask for it to be reviewed, and we will respond in writing.

Transparency. This document and the privacy notice set out which systems exist, what they use and what they affect.

Employment status. Whether you are legally an employee or self-employed is determined by the law of your country, applied to the facts of how you actually work, not by what a contract calls you. If you believe your working arrangement amounts to employment, say so and we will address it rather than rely on the label.

What we hold ourselves to

Singkatnya: Staff who operate these systems are trained on them, decisions are logged and we tell workers before a new system goes live.

AI literacy. Article 4 of the EU AI Act has applied since 2 February 2025 and requires that staff involved in operating an AI system have a sufficient understanding of it. Staff who review automated scores are briefed on what the system measures, what it cannot measure and where it is known to be unreliable.

Logging. Automated assessments and the human decisions that follow them are recorded, with the actor and the reason, so any decision can be reconstructed.

Notice before deployment. Before an automated system is put into use in a way that affects contributors, we tell affected contributors and, where they exist, their representatives.

Review. Where a system is producing results that look wrong for a group rather than an individual, that is treated as a fault in the system and not as a fact about the group.

If you think a system got it wrong

Singkatnya: Tell us, and if we do not resolve it you can go to a regulator.

Write to info@corpshore.ai describing the decision and why you believe it is wrong. A different member of staff will review it.

If you are not satisfied you may complain to your data protection authority. In the United Kingdom that is the Information Commissioner's Office; in the European Economic Area, the authority in your member state.

In New York City, complaints about automated employment decision tools go to the Department of Consumer and Worker Protection. In Illinois, the Attorney General enforces the Artificial Intelligence Video Interview Act.

Raising a complaint never affects your standing on the platform, your access to work or anything you are owed.