AI Data

Can You Withdraw Your Recordings From an AI Dataset? How Withdrawal Works

By Chisom Anen21 November 20266 min read
Can You Withdraw Your Recordings From an AI Dataset? How Withdrawal Works

Overview & Key Takeaways

This article explains how data contributors can request the withdrawal or deletion of their recordings from AI training datasets, focusing on TrustAm's policies and Nigerian data protection laws. Learn about your rights regarding the data you provide.

Key Takeaways
  • Contributors can request deletion of their approved recordings at any time.
  • Unapproved recordings are automatically deleted after 90 days.
  • TrustAm ensures data handling complies with Nigerian data protection laws.
  • Bystander faces are blurred to protect privacy.
  • Data withdrawal is a key right for individuals contributing to AI datasets.

Understanding Your Rights: Withdrawing Recordings from AI Datasets

When you contribute recordings to an AI training dataset, a common question arises: can you change your mind and withdraw that data later? The answer is yes, and understanding this process is crucial for anyone participating in data collection. TrustAm is committed to ensuring contributors have control over their data, in line with Nigerian data protection regulations.

This article focuses on how withdrawal and deletion work for our contributors and what buyers can expect regarding data lifecycle management. We aim to provide clarity on the rights and processes involved, ensuring transparency and trust in the data we collect and supply.

Nigerian woman reviewing video on her smartphone, demonstrating the control contributors have over their work.
Contributors have the right to manage their data, including withdrawal requests.

How Data Withdrawal Works at TrustAm

At TrustAm, we handle data withdrawal with clear processes designed to protect contributor privacy and comply with regulations like the Nigeria Data Protection Act (NDPA) 2023. Your ability to withdraw or request deletion of your recordings is a fundamental aspect of our service. This ensures you retain control over the information you share.

Here’s a breakdown of how it works:

  • Approved Recordings: For recordings that have passed our review and are added to our licensed dataset, you can request their deletion at any time. This process involves identifying the specific recordings associated with your contribution and initiating a formal request. We then work to remove these recordings from our active stores. This aligns with the principles of data subject rights outlined by the Nigeria Data Protection Commission (NDPC) NDPC.
  • Unapproved Recordings: Recordings that do not meet our quality standards or are flagged for any reason during the review process are not added to the licensed dataset. These unapproved recordings are automatically deleted from our systems after 90 days. This ensures that only high-quality, validated data is retained, and any rejected material is promptly purged.
  • Bystander Data: We take steps to protect the privacy of individuals who may appear incidentally in recordings. Bystander faces are blurred before any data is made available for licensing. While direct withdrawal requests apply to your own contributions, our blurring process addresses privacy concerns for others captured in the background, as detailed in our post on Face Blurring in AI Video Datasets.

It's important to understand that once data is licensed to a buyer, a full withdrawal may not always be immediately feasible due to the nature of data licensing agreements and the potential for data to be integrated into larger models. However, our policy is to honor withdrawal requests for approved recordings to the greatest extent possible, respecting your rights as a data subject.

Why Data Withdrawal Matters for AI Training

The ability to withdraw recordings is more than just a feature; it's a core component of ethical data collection and aligns with global best practices in AI development. As AI systems become more integrated into our lives, ensuring that the data used to train them is collected responsibly and with respect for individual rights is paramount. This is why regulations like the NDPA 2023 are so important for data collectors and contributors in Nigeria. Our approach to data handling is detailed in NDPA 2023 and AI Training Data.

For AI and robotics teams, understanding data provenance and lifecycle is critical. Datasets are often accompanied by documentation, such as datasheets, which outline their origin, collection methods, and any limitations. This is why TrustAm provides detailed manifests and QA reports to buyers, ensuring transparency. The concept of datasheets for datasets is discussed in detail at Datasheets for Datasets.

First-person view of a Nigerian auto mechanic working on an engine, illustrating the type of work recorded.
Workplace recordings are verified, but contributors retain rights to their data.

TrustAm's Commitment to Data Protection and Control

TrustAm operates with a strong commitment to data protection. Our processes are designed to ensure that contributors feel secure and in control of their contributions. This includes rigorous identity verification, secure storage, and clear policies on data usage and withdrawal.

Verification and Security:

  • Contributor Verification: We use BVN or NIN checks against national records, verified with a live selfie, and payments are made only to accounts in the contributor's name. This process helps prevent fraud and ensures that only legitimate individuals participate. More on this can be found in ID Checks and Selfies in AI Data Work.
  • Workplace Registration: Field agents register workplaces in person, obtaining consent from owners before any recording begins. Each workplace is documented with photos and a mapped area, ensuring legitimacy.
  • Recording Integrity: We employ multiple checks to ensure recording quality and authenticity, including challenge codes, motion sensors (IMU), GPS data, and automatic checks for issues like frozen video or duplicates. This process is detailed in Fraud Controls in Crowdsourced Video Data.
  • Secure Storage: Approved recordings are stored securely and are only accessible for licensing. Recordings that are not approved are deleted after 90 days.

This multi-layered approach ensures that the data we collect is not only high-quality but also handled ethically and securely, respecting the rights of our contributors. The EU AI Act also emphasizes data and data governance, highlighting the global importance of these practices EU AI Act, Article 10: Data and Data Governance.

Frequently Asked Questions

Can I get paid for recordings that are not approved?
No, contributors are only paid for approved minutes of recording. Recordings that do not pass our review process are not added to the dataset and do not generate payment. These unapproved recordings are also automatically deleted after 90 days.
What happens to my recordings after they are licensed to a buyer?
Approved recordings are stored for licensing. While you can request the deletion of your recordings, once data has been licensed and potentially integrated into AI models by a buyer, a complete removal might become technically challenging. We strive to honor all withdrawal requests for approved data to the best of our ability.
Are bystander faces blurred in the recordings?
Yes, bystander faces are blurred in all recordings before they are made available for licensing to protect their privacy. This is a standard procedure to ensure compliance with data protection regulations.
How long are unapproved recordings kept?
Unapproved recordings are automatically deleted from our systems after 90 days. This ensures that only quality-assured data is retained and that rejected material is promptly discarded.
What if I want to withdraw my data after it has been licensed?
You have the right to request the withdrawal or deletion of your approved recordings at any time. Please contact TrustAm support with your request, and we will initiate the process to remove your data from our active stores where technically feasible.
Do I need to pay to withdraw my data?
No, there is absolutely no charge for requesting the withdrawal or deletion of your recordings. TrustAm never asks for payment from contributors for any service, including data withdrawal requests.
Buy from TrustAm Data

Need AI training data from Nigeria?

Egocentric video of skilled work and Nigerian-language speech, collected to order. Email [email protected] or request a free 10-hour sample; a person replies within one working day.

CChisom Anen

Founder and CEO of TrustAm. Building Nigeria's reputation marketplace for verified service providers. Reviews unlock only after real payments.

Disclosure: This article is published by TrustAm Data, part of TrustAm Limited. Links to /data pages are our own data service.

Sources & References

  1. NDPC— ndpc.gov.ng
  2. Datasheets for Datasets— arxiv.org
  3. EU AI Act, Article 10: Data and Data Governance— artificialintelligenceact.eu
  4. National Artificial Intelligence Strategy (NCAIR/NITDA)— ncair.nitda.gov.ng
  5. National Information Technology Development Agency (NITDA)— nitda.gov.ng
  6. EPIC-KITCHENS— epic-kitchens.github.io
  7. LeRobot (Hugging Face)— github.com
  8. Masakhane— masakhane.io

Related Articles