© All rights reserved to Barnea Jaffa Lande Law offices
Together is powerful

Israel Securities Authority Publishes Draft Directive on AI in Investment Advice and Portfolio Management

Summary

  • On November 4, 2026, the Israel Securities Authority (ISA) published for public comment a new draft directive for licensees concerning the provision of services using technological tools. The draft updates the existing directive to address the use of artificial intelligence systems and applies the recommendations of the interministerial report, “Artificial Intelligence in the Financial Sector,” to investment advice and portfolio management.
  • The ISA has opted for a principles-based and risk-based approach. Licensees will assess the risk and materiality of each technological tool, and the scope of their obligations will follow from that assessment. The obligations include corporate governance, controls and monitoring, explainability, disclosure and outsourcing requirements, and the appointment of a “technology officer.”
  • Use classified as high-risk and highly material (“heightened use”) will be permitted only within a licensed corporation. Such use is subject to board involvement and reporting to the ISA.
  • Public comments will be accepted until December 6, 2026. The ISA is also seeking input on the use of AI agents and the integration of general-purpose AI models with financial platforms.

The Israel Securities Authority (ISA) has published the “Draft Directive for License Holders Regarding the Provision of Services Using Technological Means and Artificial Intelligence,” alongside a draft regulatory impact assessment (RIA) report. The draft seeks to implement the recommendations of the “Artificial Intelligence in the Financial Sector” report in the specific context of investment advice, investment marketing, and portfolio management.

 

The report, published in December 2025, was prepared by an interministerial team comprising representatives of the Ministry of Justice, the Ministry of Finance, the Bank of Israel, the Israel Securities Authority, the Capital Market, Insurance and Savings Authority, and the Israel Competition Authority.

 

Background: From Online Services to Artificial Intelligence

The existing directive was published in 2016. Its purpose was to adapt the provisions of the Regulation of Investment Advice, Investment Marketing and Investment Portfolio Management Law, 5755-1995 (commonly known as the Advice Law) to the provision of services online and through algorithms. It was based on the principle of technological neutrality, under which a licensee’s duties of trust and care do not change according to the tool used. The directive also provided that the licensee is responsible for harm caused to a client as a result of the use of technology.

 

According to the explanatory notes, the directive was drafted before the widespread use of AI systems. It does not fully address their unique characteristics, including a lack of explainability, model “hallucinations,” disinformation risks, and information security, cybersecurity, and privacy risks.

 

The ISA recognizes the potential of technology throughout all stages of service delivery. At the needs assessment stage, AI systems may enable analysis of client data beyond questionnaire responses. At the investment policy-setting stage, they can help screen investment options and match them to the client’s profile. During ongoing service, they may generate personalized recommendations and even automatically rebalance portfolios. They can also assist with supporting activities, such as marketing and customer service.

 

A Principles-Based Approach: What Does the Draft Include?

The RIA report examined three alternatives: maintaining the status quo, establishing detailed and rigid rules, and adopting principles-based regulation. The ISA chose the last alternative, partly because it allows flexibility in light of the pace of technological development. According to the report, this approach is also consistent with the emerging trend in the European Union, the United Kingdom, and Singapore.

 

Accordingly, the draft does not specify which technologies may be used. Instead, it requires licensees to ensure that the chosen tool enables compliance with the applicable requirements. The main elements of the proposed framework include:

 

  • A flexible definition of “AI system”: The definition was adopted from the interministerial team’s report. A licensee may determine that a computer-based system is not, in substance, an AI system. In addition, basic and commonly used systems, such as email and office software, are excluded from the directive’s scope. However, the ISA emphasizes that this is not a blanket exemption and that the determination requires an individual assessment.
  • Risk and materiality assessment: The assessment must be conducted before use begins and validated at least annually. Factors to consider include the degree of reliance on the tool, the extent of its use, the impact of a potential failure, and the level of human involvement.
  • Technology officer: This requirement also applies to an individual licensee. The technology officer need not be licensed, and an external party may be engaged for this purpose.
  • Controls, monitoring, and explainability: Controls and monitoring will apply to input data and outputs. In addition, “general explainability” of how the system operates is required. Heightened use also requires “specific explainability” of particular outputs, or compensating measures instead.
  • Disclosure to clients: A prominent notice of AI use must be provided, and misleading representations about the system’s capabilities (“AI washing”) are prohibited.
  • Outsourcing: Due diligence on providers, contractual arrangements addressing key issues, and business continuity principles are required.

 

The draft also proposes eliminating the rigid division into service stages and applying the directive even where humans are involved in the process. The provisions concerning social trading services and advice for self-directed trading will be moved to a separate directive. Until then, they will continue to apply in their current form.

 

Heightened Use: Stricter Requirements for High-Risk Uses

Use classified by the licensee as high-risk and highly material will be considered “heightened use.” The following rules will apply:

 

  • Such use may be undertaken only by a licensed corporation, not by an individual licensee.
  • Its commencement, termination, and any material change must be reported to the ISA.
  • The board of directors will be responsible, among other things, for approving the usage strategy, assessing risk and procedures, and appointing the technology officer.

 

The ISA seeks public input on two questions: whether to establish uniform quantitative or qualitative criteria for classifying heightened use—for example, criteria similar to the thresholds for a “large portfolio management company”—and whether to impose additional specific requirements on the board of directors.

 

Open Questions: AI Agents and Integration with External Models

The ISA also seeks input on whether the use of autonomous AI agents gives rise to risks not addressed by the draft.

 

In addition, the ISA raises questions about connecting clients’ financial information to external generative AI tools, including whether such a connection could constitute an unlawful referral for investment advice, contrary to Section 3A of the Advice Law. The ISA clarifies that raising this issue should not be viewed as permission to engage in this activity.

 

Implications and Preparation

The proposed effective date is six months after publication of the notice in the Official Gazette (Reshumot).

 

The draft reflects a change in approach: instead of ambiguity about the boundaries of AI use, the ISA is outlining a framework within which licensees can integrate such systems into their services. For licensees, this presents an opportunity to incorporate AI tools into client services without the regulatory uncertainty that has accompanied such steps to date. According to the explanatory notes, the ISA sees value in innovation in this field, including improving service efficiency, reducing costs, and making investment advice and portfolio management services accessible to wider audiences.

 

Realizing this opportunity, however, requires appropriate organizational preparation: mapping existing technological tools, conducting risk and materiality assessments, updating procedures and supplier agreements, and appointing a technology officer.

 

Licensees considering the implementation of AI tools, entities already making extensive use of technological tools, individual licensees, and platforms connecting clients to external AI tools are advised to examine the implications of the draft for their operating structure and consider submitting comments during the consultation period.

 

***

 

Adv. Efrat Cohen is a partner and head of the firm’s Regulation Department. 

Adv. Avihai Tal is an associate in the firm’s Regulation Department. 

 

Barnea Lande’s Regulation Department advises financial institutions and licensees on the regulatory implications of using technology and artificial intelligence. Our team assists clients in assessing the draft directive’s impact on their operations, mapping AI use cases, evaluating risks, developing policies and controls, reviewing supplier agreements, and preparing for proposed disclosure and corporate governance requirements. We also support clients in preparing comments on regulatory drafts and engaging with regulators.

Tags: AI Regulation | Artificial Intelligence

FAQs

The draft proposes a principles-based, risk-based framework for using technological tools, including AI, in investment advice, investment marketing, and portfolio management. Licensees would assess the risk and materiality of each tool and apply obligations accordingly, including governance, controls, monitoring, explainability, client disclosure, and outsourcing requirements.

The draft is relevant to licensees providing investment advice, investment marketing and portfolio management, including individual licensees and licensed corporations. It is also relevant to platforms connecting clients to external AI tools.

Use classified by the licensee as both high-risk and highly material would be considered “heightened use.” It would be permitted only within a licensed corporation and would require board involvement and reporting to the ISA when the use begins or ends, or materially changes.

The proposed effective date is six months after publication of the notice in the Official Gazette (Reshumot). Public comments on the draft will be accepted until December 6, 2026.

Barnea
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.

// load AI chatbot on test page