Machine Unlearning
Techniques that make a trained model behave as if specific data had never been in its training set without a full retrain. Exact methods (SISA, Sharded-Isolated-Sliced-Aggregated, Bourtoule et al. 2021) retrain only the affected shard and are available in production ML platforms that support them. Approximate methods statistically suppress deleted data's influence without retraining; they offer strong evidence of forgetting but not yet an ironclad mathematical guarantee, and verification remains an open research problem in 2026.
Defined in 4 GAGE programs, which carry 9 distinct definitions of it. The wording above is taught in Certified AI Practitioner: Workplace Foundations.
How each discipline defines it
The same term does different work depending on who is using it. These are the definitions as each program teaches them, unedited.
Techniques that make a trained model behave as if specific data had never been in its training set without a full retrain. Exact methods (SISA, Sharded-Isolated-Sliced-Aggregated, Bourtoule et al. 2021) retrain only the affected shard and are available in production ML platforms that support them. Approximate methods statistically suppress deleted data's influence without retraining; they offer strong evidence of forgetting but not yet an ironclad mathematical guarantee, and verification remains an open research problem in 2026.
An emerging class of techniques attempting to remove the influence of specific training examples from an already-trained model without full retraining. As of 2026, the field lacks a standardized approach and any formal, verifiable guarantee that unlearning has fully succeeded; treat claims of successful unlearning as unproven unless independently verified.
The research problem of removing a specific record's influence from a trained model. Not a routine capability, and the reason a promise of erasure from a model is unmeetable.
The emerging and, as of 2026, not-yet-reliable field of removing the influence of specific training data from an already-trained model without full retraining.
A body of research techniques attempting to remove a specific training example's influence from an already-trained model's weights without a full retrain from scratch. As of August 2026, an emerging field without standardized definitions of success or reliable, widely accepted verification methods; not a production-grade guaranteed deletion capability.
Where it is taught
The exact lessons this term appears in. The first module of every program is free with a free account.
- Privacy and Data Rights in AI Systems · Ethical and Responsible AI and Operational Governance, Certified AI Practitioner: Workplace Foundations
- Retention versus the model that memorized: deleting data a model already learned from · Consent, Purpose, and the Law of Data, Certified AI Data Governance Professional (CADGP)
- Membership inference and extraction: what a model reveals about its training data · Poison, Leaks, and the Adversary, Certified AI Data Governance Professional (CADGP)
- The deletion decision: the terabytes your organization should destroy this quarter, defended · The Money of Data, Certified AI Data Governance Professional (CADGP)
- The deletion that was not: personal data found in a model trained last year · Data Incidents, Certified AI Data Governance Professional (CADGP)
- Provenance under attack: your content credentials stripped, spoofed, and laundered, and the countermeasures that survive · Adversarial Data Governance, Certified AI Data Governance Professional (CADGP)
- Machine unlearning: the emerging answer to deleting what a model learned · The Frontier Discipline, Certified AI Data Governance Professional (CADGP)
- The retention decision: what you must keep, what you must destroy, and proving both · Data Reality, Certified AI Governance Professional (CAIGP)
- AI privacy · Privacy and Data Governance, Certified Enterprise Governance, Risk and Compliance Professional (CEGP)
Where it shows up at work
A definition is where Machine Unlearning starts. The lessons above teach it inside the competencies below, which are what job descriptions ask for and what the GAGE credential records. Every link here is drawn from the same careers map that builds a role's readiness check.
Applies GDPR, CCPA and sector rules to training data, inference, automated decisions, lawful basis, individual rights and cross-border transfer.
2 lessons on this page teach it as a primary outcome.
Builds minimization, purpose limitation, de-identification, consent and retention into an AI system before launch, with tests that prove it.
2 lessons on this page teach it as a primary outcome.
Classifies information, applies least privilege, sets retention and acceptable-use rules, and controls what may enter a prompt, a log or an embedding.
2 lessons on this page teach it as a primary outcome.
Roles that hire for it
Each role page shows the full competency set, what the seat decides, and a readiness check against your own evidence.
- Chief Privacy Officer, AI privacy editionneeds it at depth
- AI Privacy Engineerneeds it at depth · 6 open postings this week
- Privacy Analystneeds working knowledge
- AI Regulatory Counselneeds it at depth · 14 open postings this week
- Data Governance Leadneeds it at depth · 2 open postings this week
- Chief Data Officer, data leadership editionneeds working knowledge
How job descriptions say it
Postings rarely use the textbook word. These are the phrases the careers map reads as asking for the competencies above.
Where the frameworks place it: GDPR; CCPA/CPRA; EU AI Act Article 27; ISO/IEC 27701.
One interview question about it
Apply a privacy law you know to a model trained on customer records. Where is the legal basis, and where is the risk?
Asked of candidates for privacy law applied to ai. Answering it in your own words is exactly how a GAGE topic is passed, and the credential records the pass where an employer can check it.
If you come from another field
Some backgrounds already carry this competency, and the readiness check gives credit for it.
- Privacy and data protection carries it as work you can show: Privacy assessments and lawful-basis records are your work product.
- Legal and paralegal work carries it as a claim to back up: Privacy and consumer law are usually within reach.
- A law degree (JD, LLB or LLM), no AI work yet carries it as a claim to back up: Privacy and data protection law are usually a course, not yet a file.
Terms it appears with
Not an alphabetical neighbourhood: these are the terms taught in the same lessons, ranked by how often they appear together.