By Sharon Shefer, Senior Partner at Levitan, Sharon & Co.
Artificial Intelligence (AI) is the hot topic in the hi-tech industry. Every other start-up boasts about using AI and seeks to leverage vast amounts of existing data to tailor and refine the services it provides.
We are witnessing a highly charged discourse regarding the risks and dangers associated with the use of AI. We all remember the storm that the OpenAI GPT-3 unleashed upon the world (Recall? It happened just a few months ago – before October 7th…).
Like many things in our world, one way to limit and regulate risks is through legislation. So, we checked what Israeli legislation has to contribute to this, with a focus, of course, on insurance.
There is no specific legislation in the insurance world that addresses the topic of Artificial Intelligence, and, in fact, there is no legislation in Israel at all – no law, no regulations (secondary legislation), and not even guidelines from the Regulator of the Insurance and Savings Market (directives that obligate insurance companies and have legal validity).
What does exist?
Our Regulator in Israel aims to encourage Insurtech companies. This is evident in the policy paper published on 9 January 2022, by the Capital Market, Insurance, and Savings Authority (Policy Paper – Investment in Insurtech Corporations and Investment in Innovative Financial Technology Corporations). In this paper, the authority encourages local insurance companies to promote technological innovation in their respective fields and proposes a series of easements accordingly
A similar approach can be seen in its attitude towards Artificial Intelligence, as will be apparent later on.
Artificial Intelligence – Ministry of Innovation – October 2022
In October 2022, the Ministry of Innovation published a draft of “Regulation and Ethics Principles in the Field of Artificial Intelligence,” aiming to stimulate discussion on the subject.
The document addresses concerns about this technology, such as privacy infringement, cyber-attacks, improper bias in decision-making, etc.
The document seeks to promote the following principles:
- Responsible innovation – balancing the need to protect individuals from challenges with the interest in advancing this technology.
- Artificial intelligence as a catalyst for growth.
- Human-centric approach – emphasizing human rights and public interests, ensuring that the development and use of AI comply with the law, fundamental rights, and public interests while preserving human dignity and privacy.
- Equality and prevention of unacceptable bias – considering the need for equality, diversity, concerns about bias in AI systems, and the risk of unacceptable discrimination against individuals or groups.
- Transparency and accountability – the need to inform those in contact with AI or affected by its operation and providing explanations for its decisions or how it operates.
- Reliability, security, and safety.
The document clarifies that future legislation on this subject will be what is referred to as “soft” legislation, one that does not intervene and enforce its principles, and primarily sectoral, aligning with existing laws based on the principles mentioned above, rather than comprehensive legislation addressing the entire field.
The legislation will focus on safeguarding the rights of those affected by decisions of Artificial Intelligence and imposing various obligations on key users of algorithmic systems.
Apart from “standard” legislation, the document also discusses establishing ethical principles for the field of Artificial Intelligence, Regulator recommendations for voluntary adoption, and self-regulation.
Industries in which it is already clear that they will be affected: intellectual property rights, competition, privacy and data security, defamation, and cyber-security.
Government Decision 173 from February 2023, titled “Strengthening Israel’s Technological Leadership,” approves the national plan for Artificial Intelligence and sets several goals, including accelerating basic and applied research in the field, creating a leap for the Israeli industry in developing infrastructure for AI, and implementing AI applications in the public sector.
The Ministry of Finance will allocate a budget for this purpose, with the total investment outlined in the plan amounting to one billion shekels in two stages.
In December 2022, an inter-ministerial team began examining the use of Artificial Intelligence in the financial sector. On 29 March 2023, a public call was published inviting interested entities in the field to provide information to the inter-ministerial team (Ministry of Finance, Ministry of Justice, Israel Securities Authority, and Others). This reflects the anticipation of the need for legislative arrangements regarding the subject.
Since that announcement, there has not been much progress in Israeli legislation of this matter.
Europe
The expected leap came from Europe, which aims to lead legislation on this subject. Here, it is not about soft legislation but rather about the regulation of the field. In June, a comprehensive draft legislation was published, including distinctions between different levels of risk posed by AI systems.
Highly risky systems for humans – systems using manipulative or subliminal messages, exploiting humans, or used for social ranking based on economic or behavioral status – will be prohibited
Several other uses are proposed to be restricted, such as:
– Biometric remote identification in real-time in public places
– Biometric remote identification for law enforcement (only in serious crimes and with judicial approval)
– Biometric sorting based on characteristics such as gender, ethnicity, race, or political stance
– Emotional sentiment identification for law enforcement, workplace, and educational institutions, etc.
High-risk systems are defined as those causing harm to health, safety, fundamental rights, etc.
Providers of large language model systems (LLMS) such as OPENAI, Microsoft, Google, or Meta will be subject to comprehensive transparency requirements.
- Tools that generate deepfake images or videos will be required to label the products to clarify that they were created by AI.
- Facial recognition technologies will only be limited to cases related to safety or national security.
- Significant fines are expected.
Companies developing tools will need to provide regulators with risk assessments and details of the information used for model training. They will also need to demonstrate that the system does not cause harm, such as bias based on race. Open-source AI models will be exempt from some restrictions, based on the understanding that if the code is available for viewing, analysis, and modification by any developer worldwide, the general public can better handle issues than regulatory agencies.
In the United States, in October 2022, a proposal for an AI Bill of Rights was developed based on five principles:
– Safe and effective systems
– Protection against algorithmic discrimination
– Data protection
– Notification and explanation
– Human alternatives
The topic is not fading away, and discussions are ongoing in the Senate and Congress, calling for a legislative series on the subject
The crucial aspect in our view is that Insurtech companies serve as the gateway/opening for insurance companies to enter the world of Artificial Intelligence. Through them, AI will be integrated into the systems of insurance companies, and regulation must be closely monitored. Undoubtedly, it is expected to emerge in this world and is already advancing, particularly in Europe and the United States, and is anticipated to take effect here as well.
