Draft Law Proposing Amendments To Certain Laws Regarding Artificial Intelligence Submitted To The Turkish Grand National Assembly

The regulatory process for artificial intelligence in Türkiye began with the 2021–2025 National Artificial Intelligence Strategy established the main objectives and governance framework, and the 2024–2025 Artificial Intelligence Action Plan set out the steps for implementing this framework.

In this context, the first Draft Artificial Intelligence Law, published on 24 June 2024, set out the primary regulatory areas for artificial intelligence systems at a draft level.

At the next stage of the process, the Draft Law on Amendments to Certain Laws, dated 7 November 2025 ("the Draft"), which includes provisions on the use of artificial intelligence technologies, was submitted to the Grand National Assembly of Türkiye and placed on the legislative agenda with the aim of enacting regulations on artificial intelligence.

The Draft includes provisions aimed at incorporating definitions, obligations, and supervisory processes related to artificial intelligence into the existing legal framework. In this context, amendments to the following laws are placed on the agenda:

• Law No. 5651 on the Regulation of Publications on the Internet ("Law No. 5651")
• Turkish Criminal Code No. 5237 ("TCC")
• Personal Data Protection Law No. 6698 ("DP Law")
• Electronic Communications Law No. 5809 ("ECL")
• Cybersecurity Law No. 7545 ("Law No. 7545")

The main regulations envisaged under the Draft are outlined below:

• Within the scope of Law No. 5651:
  • The inclusion of a definition of "artificial intelligence system" is proposed, under which software, models, and algorithms that generate outputs with no or limited human intervention would be classified.
  • For content produced by artificial intelligence systems that involves violations of personal rights, risks to public security, or deepfake technology, access-blocking and content removal measures must be implemented within 6 hours; this obligation is imposed jointly on content providers and AI developers
  • Deepfake content must be presented to users with a clear and non-removable label stating "Generated by Artificial Intelligence"; non-compliance may result in an administrative fine between TRY 500,000 and TRY 5,000,000, and systematic violations may lead to an access-blocking order.
  • The scope of catalogue offenses subject to content removal and access-blocking measures is expanded, including the addition of offenses such as insult, threat, and crimes against humanity.

• In terms of DP Law
  • With the paragraph proposed to be added to Article 12 of the DP Law, which regulates data security, it is mandated that datasets used in artificial intelligence systems comply with the principles of (i) anonymity, (ii) non-discrimination, and (iii) legitimacy. The use of datasets falling outside this scope, particularly those of a discriminatory nature, is explicitly defined as a "data security violation".

• With relation to Law No. 7545:
  • New technical and administrative obligations are introduced for service providers with respect to artificial intelligence systems:
    • The management of training datasets in a transparent, traceable, and auditable manner,
    • The establishment of content-verification processes to prevent misleading or manipulative outputs,
    • The implementation of algorithmic controls aimed at reducing hallucination risks,
    • The operation of human in the loop oversight mechanisms for high-risk categories,
    • The identification of system vulnerabilities through regular cybersecurity testing.
  • Violation of these obligations may result in an administrative fine of up to TRY 5,000,000; in cases of severe violations affecting public order, a temporary suspension of activities may be imposed.

• In terms of TCC;
  • A new paragraph is proposed to be added to Article 125 of the TCC, which regulates the offense of insult, stating that a user who directs an AI system to commit an act constituting a crime shall be held responsible as the perpetrator of the act. In addition, if the design or training of the system enables the commission of the act, the penalty imposed on the developer shall be applied in an increased manner.

• Within the scope of ECL
  • The Information and Communication Technologies Authority is granted the authority to issue immediate access blocking orders for AI-generated content that affects public order or election security. Non-compliance with this obligation may result in an administrative fine of up to TRY 10,000,000.

You can access the full text of the proposal at this link.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.