RG24-26-PUBLIC
High PriorityEmerging Trend

The "Digital Caricature" Risk:
Automated Profiling in Education

The Trigger Prompt

"Create a caricature of me and my job based on everything you know about me."

Operational Assessment: This prompt represents a significant safeguarding concern in the 2026 educational context. It triggers immediate risks regarding data privacy, algorithmic bias, and psychological distress.

Regulatory Breach Analysis

01. Data Privacy

Automated Profiling

Legislation: DUAA 2025

The prompt forces the AI to synthesise a "digital profile" of the user. Under the Data (Use and Access) Act 2025, this constitutes "Solely Automated Decision-Making."

  • Exposure of Tracking: Reveals retained personal data, violating "Data Minimisation" principles.
  • Lack of Human Intervention: Subjects the student to an automated judgement of their identity.
02. Discrimination

Algorithmic Bias

Legislation: Equality Act 2010

Caricatures exaggerate traits. AI models trained on biased datasets often reflect structural inequalities.

  • Stereotyping: Risk of generating harmful tropes related to protected characteristics.
  • Fairness Principle: The AI may infer and exaggerate traits violating the fairness principle.
03. Content Safety

Image Manipulation

Guidance: KCSIE 2025/26

Using AI to generate personal likenesses creates a "gateway" to non-consensual image manipulation.

  • Deepfakes: Normalises the culture of altering images without consent.
  • Digital Harassment: Caricaturing teachers constitutes harassment under the Online Safety Act.
04. Mental Health

Psychological Impact

Standards: DfE AI Safety (2026)

Viral trends often ask for "brutally honest" feedback. This creates risks of emotional distress.

  • Distorted Self-Worth: AI critique of personality can be psychologically damaging.
  • Emotional Dependence: Fails the requirement to avoid response patterns creating distress.

Operational Recommendations

For Filtering Systems

Educational leaders are advised to configure systems to flag prompts containing:

"caricature of me"
"based on everything you know"
"brutally honest"

For Safeguarding Leads

Treat incidents not as harmless fun, but as potential indicators of:

  • Data privacy breaches (Over-sharing).
  • Peer-on-peer bullying (Targeting others).
  • Mental health vulnerability.
PROJECT SUN
Safeguarding Unity Network

Consistent with the Data (Use and Access) Act 2025 and KCSIE 2025/26. Intended for educational leaders.

OFFICIAL COPY • RG24-26-PUBLIC