New Innovation Briefing about Cyber Insurance-Models and Methods and the Use of AI by European Union Agency for Cybersecurity (ENISA)

Data Protection Law

On 21st February of 2024, the European Union Agency For Cybersecurity (ENISA) released a new report which introduces cyber risk and cyber insurance, provides an overview of existing research and modelling approaches, and identifies gaps.

The report, which consists of 62 pages in all, analyses the tasks of the cyber security industry, data breaches, types of cyber risks, machine learning (ML), and artificial intelligence (AI) methods.

The study is a further development of the report prepared by ENISA in 2023 analyzing the current perspectives and challenges faced by operators of essential services (OESs)1 in relation to the subscription of cyber insurance services, with practice recommendations.

After being defined as "Insurance companies are confronted with complex tasks when assessing and managing risks, which must also be adequately implemented in the area of cyber insurance" for insurance companies, the tasks of cyber-insurance industries are listed as follows:

  • pricing,
  • insurance-portfolio management,
  • reserving,
  • reinsurance,
  • preventing future losses.

This research also includes data sources that provide initial insights into cyber claims and interesting data to be used in research.

 

The following assessments in the research are of crucial relevance:

 

  • Known data sources on data breaches are the `Chronology of Data Breaches’10, organized by the nonprofit corporation `Privacy Rights Clearinghouse’ (PRC) and the `Open security foundation data loss data-base’.
  • Compared to academic research, the data situation is slightly better for insurance companies, as they have additional non-public information at their disposal, e.g. collected via risk questionnaires on the companies in their portfolio.
  • Cyber risk is a (relatively) new field and the relevant data are scarce.
  • It is clear that one of the biggest obstacles to ML/AI research and progress in cyber risk and cyber insurance is the lack of available data. Therefore, ENISA strongly supports the vision of creating a public repository on cyber security and cyber insurance.

 

Cyber risk types are identified as below:

  • Idiosyncratic risks: Idiosyncratic risks refer to independent fluctuations/losses at the level of individual assets. In homogenous pools, these are classical risks in the insurance business.
  • Systematic risks: Systematic risks refer to the functional dependence of cyber risks on underlying random processes, such as the random development of security or attack technology or information flows.
  • Systemic risks: Systemic risks refer to local and global feedback and interaction effects that cannot be explained solely on the basis of idiosyncratic fluctuations and the impact of external processes on individual assets.

 

n addition, ENISA believes that ML/AI also offers interesting use cases for cyber insurance. However, in the context of cyber insurance, ENISA needs to pay particular attention to the following issues:

 

  • Difficult access to data for academic researchers and data scarcity in general.
  • The specific stylized facts of cyber data (accumulation risk, non-linear dependencies, non-stationarity, ...) may require a different adoption of ML/AI methods.
  • An interdisciplinary approach is required, e.g. actuarial issues, legal opinion, IT knowledge

 

etc. have to be combined.

 

Finally, the briefing analyzed 23 research topics in detail under the headings of objectives, formations, scope, innovation opportunities, expected results, and outputs.

You can reach further information here.

 

Kind regards,

Zumbul Attorneys-at-Law

info@zumbul.av.tr