Copyright Law (2026), copying to develop artificial intelligence products and algorithms
In force since .
A copyright and text and data mining (TDM) rule binding public and private bodies.
- How the hook was established
- express
- What makes it apply
- Content origin
- Audit expectation
- none
What it requires
- Copy a work in Saudi Arabia to develop artificial intelligence products or algorithms only where the work was lawfully published, the copy was lawfully acquired and the copying goes no further than that purpose needs, and keep records of the type, source, purpose and date of each use to give to a competent authority hearing a dispute.
- Do not republish, distribute, make available or unnecessarily embed a work used for AI development in a final product without the rights holder's permission unless the work is in the public domain, and do not use it in a purely commercial framework unless the use is non-essential to the work or does not affect its normal exploitation.
These lines are LexLint's reading of the law's text.
In the frameworks
- NIST AI RMF
- MAP 4.1Approaches for mapping AI technology and legal risks of its components – including the use of third-party data or software – are in place, followed, and documented, as are risks of infringement of a third party’s intellectual property or other rights.
Our reading of where this law's duties sit in each framework. It never says that running a control meets the law. How we read a framework against the law
What record it asks for
Who may demand the log
Court
What the log must hold
Event time, Input reference, Data lineage
Logging duty
Implementing Regulation Article 30(3) requires the party that copies a work to develop artificial intelligence products or algorithms under Article 26(4) of the Law to keep records of the type of work used, its source, the purpose of the use and the date of the use, and to give them on request to any competent authority hearing a dispute about that use. The text sets no retention period and no regular audit, and the record is reachable only once a dispute is before an authority, so the audit expectation is none and access is coded to the dispute-hearing body as a court. The Law's own scope clause (Article 2) reaches works of Saudi or resident authors and works first published in the Kingdom, and that content-origin test is the hook.
- Kind
- Explicit
- As of
- Provision
- Implementing Regulation Art. 30(3), under Law Art. 26(4)
- Trigger
- crawl_access
What this law does
Article 26(4) lets a person copy the original of a work, without the author's permission and without compensation, to develop artificial intelligence products and algorithms, provided the work was lawfully published, the copy was lawfully acquired and the copying stays within what the purpose needs. Article 37(1) requires every use under Articles 26 to 36 not to conflict with the normal exploitation of the work or unjustifiably prejudice the legitimate interests of rights holders.
The Implementing Regulation requires the copying and analysis not to extend to republishing, distributing or directly exploiting the work commercially. It bars use of the work in a purely commercial framework unless the use is non-essential to the work or does not affect its normal exploitation. It requires the developer to keep records of the type, source, purpose and date of each use and to give them to any competent authority hearing a dispute.
It bars altering, republishing, making available or unnecessarily embedding the work in final products without the rights holder's permission unless the work is in the public domain. Under Article 60 of the Law the Regulation applies from the Law's own commencement.
When LexLint raises it
When your app profile says your app crawls the web or trains models.
Back to the example · Lint your app
Page updated .