AI has swept the world by storm, and, as with any technological innovation, how it will evolve and be leveraged by businesses in not only the coming years, but the coming months, is unclear.
The technology is advancing rapidly, and the benefits for businesses are undeniable, from improved efficiency and automation, to better customer service, to increased productivity, to enhanced analytics... just to name a few.
But the benefits come at a price: AI is the Wild West, and, as such, it carries risks.
AI lacks regulations. There are only loose guidelines and ethics provided for industries that individual businesses can take…or leave…at their discretion. While it’s difficult to know where, exactly, AI is going, we do know that AI poses cybersecurity risk, regulatory and compliance risk, misuse and error risk, plagiarism and intellectual property risk and risks associated with transparency and bias. While strategies and tactics do exist to address each risk, the question remains: Are they enough?
Take Amazon. Its AI-based experimental recruiting tool was found to be biased against women when reviewing resumes. Because most resumes were submitted by men, the AI was “taught” to favor male candidates–even when specific women were far better qualified.
Amazon disbanded the program, but it highlights how AI lacks contextual understanding and leaves the door open to unintended discrimination. This leaves businesses vulnerable to not only legal consequences but reputation damage.
To avoid this risk, businesses could ensure they have employment practices liability insurance (EPLI). However, EPLI policies have limitations, and coverage specifics can vary among insurance providers. Certain exclusions and conditions may apply, too.
See also: The Risks of AI and Machine Learning
The Challenge With Insuring Against AI Risk
Like EPLI, numerous insurance options exist within the traditional insurance market to address AI risk. For example, cyber liability insurance addresses losses associated with data breaches or cyber attacks, because AI systems often process large amounts of data. There’s also technology errors & omission (E&O insurance) for claims related to AI system failures or glitches, and directors and officers (D&O insurance) can protect company leadership from AI-related oversights. Intellectual property (IP insurance) provides protection from claims stemming from copyright or intellectual property infringement. Insurance to address AI risk is available, but with each of the policies mentioned, the following challenges exist:
- Complexity: AI poses complex risks. Technological innovation and advancements, like AI, are new risks that may be difficult to understand and predict. Rapid advancements make it challenging to anticipate the full extent and impact. Insurers may face uncertainty in defining policy terms, exclusions and coverage limits, leading to ambiguity in insurance contracts.
- Lack of historical data: Insurance is traditionally based on actuarial data and historical patterns of risk. However, emerging technologies and evolving risks often lack sufficient historical data, making it challenging for insurers to accurately assess and quantify these risks.
- Lack of standards: With evolving risks, there may be a lack of industry standards and best practices. This makes it difficult for insurers to establish consistent risk assessment methodologies and underwriting criteria. This can lead to inconsistencies in coverage offerings across different insurers.
- Shifting legal and regulatory implications: Technological advancements often outpace the development of regulations. The absence of established guidelines can make it challenging to determine liability and coverage parameters for emerging risks.
In addition to the challenges above, if we go back to the EPLI example and the fact that limitations and exclusions create coverage gaps and vulnerabilities, any of the insurance options mentioned have the same downsides. You can secure cyber liability insurance to protect against AI risk, but if the claim is based on losses stemming from misuse tied to human error, the claim will likely be denied.
Alternative Solutions to Insuring Against AI Risk
If so many problems exist with insuring against AI risk, does this mean insurance is useless? The answer is a clear “no.” Traditional insurance continues to be necessary, but it’s critical that businesses consider solutions to fill the gaps and limitations in these policies–or else be left to costly exposures.
One solution that businesses are using is cyber insurance. When cybercrime or a cybersecurity issue causes business interruption or a breach, a cyber policy will likely cover it. However, AI is complicated. Harvard Business Review recently published an article, “The Case for AI Insurance,” that said businesses are “woefully unprepared” for AI risk because cyber policies often won’t cover AI failures resulting in brand damage, bodily harm and property damage. Also, AI-related risk continues to be an active and evolving area of research, according to the article.
In terms of an actual, AI-specific policy, that’s yet to be determined due to market variables as the technology is still the Wild West in terms of regulations, the legal landscape and actuarial data. However, a captive insurance company could write an AI policy that addresses everything AI-risk-related the business faces.
Captive insurance is a licensed insurance company owned by the business or business owner that provides insurance coverage exclusively for the parent company. It allows businesses to manage their risks directly and more cost-effectively. For AI risks, captive insurance is especially useful because businesses can customize their policies to align with their specific AI initiatives. It offers greater flexibility in terms of coverage, underwriting and claims handling, empowering businesses to address the unique and evolving AI risk landscape. A policy written by a captive insurance company can remove exclusions a traditional insurance policy might have.
Also, captive insurance policies are particularly useful to address risks that are difficult to insure against because the premiums paid to the captive insurance company, minus claims, are retained as profit. This profit accumulates and can then be used to cover lost revenue or litigation fees tied to AI failures or issues.
See also: The Rise of AI: a Double-Edged Sword
In Conclusion
Traditional insurance, AI insurance and captive insurance can all play a role in effective and comprehensive AI-related coverage. Traditional insurance is ideal for covering broader risks, while AI coverage and captive insurance provide more specialized coverage. Beyond insurance, businesses should also stay informed regarding AI developments, best practices and emerging regulations–establishing a robust approach to risk management.
