Korea AI Training Data Copyright Damages Calculator

Estimate Korea AI training-data copyright damages from work count, license value, AI service revenue, training method, opt-out status, and output similarity.

AI Training and Output Inputs

Total mid damages

₩259,143,750,000

Total range

₩129,763,125,000 - ₩431,555,625,000

Recommended method

statutory

Litigation cost range

₩13,872,805,625 - ₩39,795,180,625

AI Copyright Factors

Per-work best estimate
₩1,297,631,250 - ₩4,315,556,250
Consolation money
₩9,562,500
Similarity contribution rate
67.6%
Aggravation rate
55%
Mitigation rate
20%
Litigation economy
Yes

Related calculators

What Is the Korea AI Training Data Copyright Damages Calculator?

This calculator estimates damages when copyrighted works are allegedly used in AI training data or when AI output resembles protected works. It covers text, images, music, video, software, datasets, AI service scale, training method, opt-out status, and output similarity.

The English version calls the Korean AI training damages function. It preserves the contribution-rate model, license-based method, AI-provider-profit method, statutory damages, comprehensive consolation-money method, and litigation-cost analysis.

Important Note

AI copyright law remains fact-sensitive and rapidly developing. The calculator is a Korean damages planning model for evidence, negotiation, and litigation economics, not a prediction of a final court ruling.

Training Stage and Output Stage

StageTypical factsCalculator inputs
Training-stage useWeb crawling, API scraping, direct collection, third-party datasetsTraining method, data scale, public dataset status, opt-out status
Output-stage similarityGenerated text, image, music, video, or code resembles a protected workOutput samples, similarity score, commercial use, output duration
Service economicsFree, freemium, or paid AI service with estimated revenue and profitService type, scale, MAU, revenue, profit rate
Rights-holder contextIndividual creator, media company, famous artist, registration statusWork count, market value, license fee, annual license income

Damage Methods

  • AI-provider-profit damages estimate the share of AI service profit attributable to the protected works.
  • License-based damages use normal licensing fees, output frequency, similarity, and contribution.
  • Statutory damages reflect the Korean copyright statutory-damage framework when actual loss is difficult to prove.
  • Comprehensive damages add consolation money to the strongest economic method for personal or reputational harm.
  • The result multiplies the selected per-work estimate by the number of affected works and then calculates litigation cost.

Evidence Package

The Korean content stresses preserving AI outputs with prompt, date, time, model name, account information, and comparison to the original work. If training-data inclusion can be checked through public tools or dataset records, those records should be saved before they change.

A rights holder will usually need creation records, publication history, license fee data, opt-out notices, service terms, similarity comparison, and evidence of commercial use or substitution.

FAQ

Why does similarity score matter so much?

The Korean model converts similarity into a contribution rate. Higher similarity can move the case from weak resemblance toward direct copying or substantial similarity.

Does opt-out always control the legal result?

No. Opt-out status is one factor. Ignoring a clear opt-out can aggravate the claim, but liability still depends on training facts, output facts, license scope, and legal exceptions.

Legal Use Notice

This English page translates and adapts the Korean calculator content for Korea-based estimates. The interactive calculator reuses the same Korean pure calculation function as the Korean page, so the numerical result is aligned with the Korean service. It is an estimation tool, not legal advice for a specific dispute.