Do Not Train AI with Your Students' Data (Please)

Training AI with student data through free personal accounts is a severe LGPD compliance risk that exposes sensitive minor data to public training models without a Data Processing Agreement. For educational institutions, avoiding this liability requires deploying enterprise AI within a Virtual Private Cloud to guarantee data isolation and auditability.

TL;DR Asking teachers to use free ChatGPT with student data is an LGPD incident waiting to happen. A minor's data is sensitive by default. Solution: AI running in a Google Cloud VPC with a DPA contract, "Your data, your model" — nothing becomes public training data, everything is auditable by the ANPD.

Innovation without governance is legal risk. Full stop. In the rush to use AI, many higher education institutions are asking teachers and students to use personal ChatGPT accounts.

This is an LGPD disaster waiting to happen. And when it does, the liability falls on the institution — not on the teacher who was "just trying to help".

Personal ChatGPT vs Gemini Enterprise in VPC

Free personal account

ChatGPT / Gemini / Copilot free

Data may be used to train a public model
No DPA (data processing agreement) with vendor
No auditable LGPD legal basis
History in the teacher's account — leaves with them if they resign
Nothing acceptable in an ANPD inspection

Enterprise in VPC

Gemini Enterprise / Vertex AI

Contractual: data never trains a public model
Signed DPA, sub-processors listed
Encryption in transit and at rest
Access logs retained by the institution
Data residency in sa-east1 (SP)

Your Data, Your Model

At Autenticare, we follow the mantra "Your Data, Your Model". When we implement AI at your institution, it runs inside a VPC (Virtual Private Cloud) on Google Cloud — isolated, audited, and under your control.

⚠️ Minors = sensitive data by default Art. 14 of the LGPD requires a specific legal basis for children's and adolescents' data — specific parental consent, not the generic enrollment contract. Before go-live: DPIA signed by DPO, notice to guardians, public AI policy published on the website, accessible opt-out channel. ANPD inspections of higher education institutions are increasing — there is no room for improvisation.

Handing over your institution's intellectual property — lesson plans, grading, student history — to train a third party's algorithm is not smart strategy. It is negligence disguised as innovation.

Frequently Asked Questions sobre Do Not Train AI with Your Students’ Data (Please)

What are the risks of using the free ChatGPT with student data? Using the free ChatGPT with student data can lead to LGPD incidents, especially since data of minors is considered sensitive. The data can be used to train the public model, there is no DPA (data processing agreement), and there is no auditable LGPD legal basis.

What is a safe alternative for using AI in educational institutions? A safe alternative is to use an AI running in a VPC (Virtual Private Cloud) on Google Cloud with a DPA contract. This ensures that the data will not be used for public training, and everything is auditable by the ANPD.

What does ‘Your Data, Your Model’ mean? ‘Your Data, Your Model’ means that the AI runs within a VPC (Virtual Private Cloud) on Google Cloud, isolated, audited, and under the institution’s control.

What are the LGPD requirements for data of minors? Art. 14 of the LGPD requires a specific legal basis for data of children and adolescents, including specific parental consent. It is also necessary to have a RIPD signed by a DPO, notice to those responsible, a public AI policy published on the website, and an accessible exclusion channel.

LGPD Governance for Higher Education

Would your institution pass an ANPD inspection tomorrow?

Autenticare delivers: Google Cloud VPC in sa-east1, DPA with sub-processors, DPIA focused on minors, public AI policy, Workspace for Education with SSO, teacher and DPO training. No surprises.

Schedule assessment → Why Gemini vs ChatGPT