Back to the MIT repository
6. Socioeconomic and Environmental2 - Post-deployment

Undermining creative economies

LMs may generate content that is not strictly in violation of copyright but harms artists by capitalising on their ideas, in ways that would be time-intensive or costly to do using human labour. Deployed at scale, this may undermine the profitability of creative or innovative work.

Source: MIT AI Risk Repositorymit256

ENTITY

2 - AI

INTENT

2 - Unintentional

TIMING

2 - Post-deployment

Risk ID

mit256

Domain lineage

6. Socioeconomic and Environmental

262 mapped risks

6.3 > Economic and cultural devaluation of human effort

Mitigation strategy

- Establish a mandatory, collective rights management system for Large Language Model (LLM) training data, utilizing micropayments or a standardized licensing framework to provide creators with equitable remuneration when their copyrighted or commercially substitutable works are ingested, thereby mitigating the systemic economic devaluation of creative labor. - Develop and implement auditable provenance tracking mechanisms for LLM outputs, coupled with technical safeguards (e.g., *k*-nearest neighbor checks) to actively flag and suppress content that poses a high risk of being a non-infringing yet highly substitutable derivative of existing creative work, requiring human review before deployment. - Mandate the adoption of enterprise-grade LLM licenses or internal AI policies that clearly define the scope of use, establish strict indemnification clauses against economic harm arising from non-infringing substitution, and require comprehensive documentation of human creative contribution for all AI-assisted works.

ADDITIONAL EVIDENCE

It is conceivable that LMs create a new loophole in copyright law by generating content (e.g. text or song melodies) that is sufficiently distinct from an original work not to constitute a copyright violation, but sufficiently similar to the original to serve as a substitute, analogous to ‘patent-busting’ (Rimmer, 2013). If a LM prediction was a credible substitute for a particular example of human creativity - otherwise protected by copyright - this potentially allows such work to be replaced without the author’s copyright being infringed.