Elicitation of security threats and vulnerabilities in Insurance chatbots using STRIDE Scientific Reports

A Cybersecurity Expert Warns Against Oversharing to AI Chatbots

The intersection of insurance with cutting-edge technologies such as generative AI, blockchain, and the IoT is reshaping how insurance products are designed, priced, and delivered. CAPE Analytics combinesdata science and computer vision to provide in-depth assessments of over 100 million properties. CAPE AIRE’s property analysis factors in a range of variables, such as how far structures are from bodies of water and highways. As a result, firms can make more informed decisions when underwriting insurance policies for, trading and investing in properties.

The nonsignificant impact of PEOU on PU does not contradict the relevant positive link between PEOU and attitude and BI mediated by PU reported in the literature. In the field of insurance new tech products, Huang et al. (2019) and de Andrés-Sánchez and González-Vila Puchades (2023) found this. Basically, two chatbots have the awkward first conversation about «So, what are your hobbies?» Then, you can review the chat logs and see if you think the other real person is worth actually messaging in real life.

Sampling protocol, respondent profile, and questionnaire

High inbound message volumes and rising customer care standards have left support teams hustling to keep resolution times low. Botshit is one example of how the use of AI might worsen companies’ customer service. As businesses employ generative AI, the researchers said employees must be more critical in managing chatbot-generated responses. Figure 9 depicts when the user has already been given rights to access the Claims chatbot. Then, the user requests information and asks FAQ (frequently asked questions) related to the claim.

Babylon Health’s platform leverages an AI-powered chatbot to generate diagnoses based on user responses. Users can interact with the chatbot in the same way they would when talking to primary care providers or other health professionals. Fishbein and Ajzen (1975) provide a commonly agreed upon definition of ATT as a “learned predisposition to respond in a consistently favorable or unfavorable manner with respect to a given object”.

Latest Articles

By detecting anomalies and patterns indicative of cyber threats, these tools can provide early warnings and recommendations for mitigating risks. For example, Beazley’s cyber risk assessment platform uses machine learning to analyse clients’ IT infrastructure and identify vulnerabilities, helping businesses strengthen their cyber defences. Automated underwriting can also improve customer experiences by reducing the time required to issue policies.

After working with IBM for three years to leverage AI to take drive-thru orders, McDonald’s called the whole thing off in June 2024. A slew of social media videos showing confused and frustrated customers trying to get the AI to understand their orders. Restaurant chains across the US are increasingly relying on AI to take orders and eventually reduce the amount of human labor needed. Among those chains are McDonald’s, Taco Bell, Popeyes, and Wingstop, all of which are using AI-powered voice bots to take orders at the drive-thru and over the phone. In its most recent cost-cutting measures in November, the company laid off roughly 17% of its global full-time employees as well as reducing its monthly expenditures.

• The results show that the low acceptance of chatbots can be explained by the use of TAM constructs, performance expectancy and ease expectation along with trust.
• Insurtech refers to the use of technology to automate and enhance processes in order to cut costs  and improve efficiency in the current insurance industry model.
• Answering minor health queries like that also frees up healthcare professionals to spend more time on their core activities.
• Paypal also offers an AI-powered customer service chatbot, and it can run through Facebook messenger.
• CEO of INZMO, a Berlin-based insurtech for the rental sector & a top 10 European insurtech driving change in digital insurance in 2023.

The study found that the algorithm used healthcare spending as a proxy for determining an individual’s healthcare need. But according to Scientific American, the healthcare costs of sicker Black patients were on par with the costs of healthier white people, which meant they received lower risk scores even when their need was greater. Zillow Offers was a program through which the company made cash offers on properties based on a “Zestimate” of home values derived from an ML algorithm. But a Zillow spokesperson told CNN the algorithm had a median error rate of 1.9%, and could be as high as 6.9% for off-market homes. Moffatt took Air Canada to a tribunal in Canada, claiming the airline was negligent and misrepresented information via its virtual assistant.

However, a limitation of this study is that our findings are based on a single case study, which limits the scope of generalisation of our results. Nevertheless, this study is a worthy exploration of the data security of insurance chatbots, which has received very little attention thus far in the literature. A current initiative by IBM involves collecting publicly available data relevant to property insurance underwriting and claims investigation to enhance foundation models in the IBM® watsonx™ AI and data platform. The results can then be used by our clients, who can incorporate their proprietary experience data to further refine the models. These models and proprietary data will be hosted within a secure IBM Cloud® environment, specifically designed to meet regulatory industry compliance requirements for hyperscalers. The risk management solution aims to significantly speed up risk evaluation and decision-making processes while improving decision quality.

The AI-based app can also recommend products to buyers that they might want through the use of personalized prompts. Microsoft’s Bing was one of the first companies to collaborate with OpenAI regarding ChatGPT. Before Google announced the release of Bard, Bing unveiled its search engine powered by ChatGPT in early February 2023. Bain & Company has announced a global collaboration with OpenAI, working to bring ChatGPT’s capabilities to clients around the world. When I tried it, the Volar bot made up some stuff about me —like that I liked pineapple on pizza (which I’ve never tried) — in a conversation about food with the other AI bot. Very minor detail, but doesn’t really give me great confidence that the app will match virtual me with the perfect virtual someone else.

Although Wendy’s CEO Todd Penegor said the chatbot won’t replace its human workers, chief information officer Kevin Vasconi said the chatbot is «as good as our best customer service representative, and it’s probably on average better.» Wendy’s is one such company that is automating its drive-thru service and training its chatbot to recognize customer lingo when taking orders and suggesting better offers. A spokesperson at the financial services company confirmed to Forbes that it is «imposing usage limits» on the chatbot and is exploring «safe and effective» ways to deploy the tech across the company.

The findings of this paper have fair potential theoretical and practical implications. They outline the special relevance of trust in explaining customers’ acceptance of chatbots since this construct impacts directly on attitude but also perceived usefulness and perceived ease of use. Likewise, improvements in the utility and ease of use of robots are also needed to prevent customers’ reluctance toward their services.

Over time, companies that continue to invest in tech advancements and machine learning for chatbot deployment will eliminate repetitive and time-consuming tasks, while also cutting costs. The chatbot offers patients 24/7 access to care, and pairs users with specific healthcare providers for virtual consultations. In August 2019, the chatbot achieved unicorn status – allowing it to surge ahead with an aggressive expansion plan. Lemonade’s policy chatbot, Maya, can onboard customers in as little as 90 seconds, compared to the approximately 10 minutes it would take with traditional insurers online.

There are thousands of chatbots in use today, on websites, messaging apps, and social platforms. The Insurtech sector is set for continued innovation and growth as it leverages advanced technologies to meet evolving consumer demands. The trends highlighted, from embedded insurance to AI-powered chatbots and blockchain, illustrate the industry’s dynamic nature and its commitment to enhancing efficiency and customer experience. Blockchain also enhances security and reduces fraud by providing a tamper-proof record of transactions. Every transaction recorded on a blockchain is immutable and transparent, making it difficult for fraudsters to manipulate data or submit false claims.

Flo Chatbot

Drawing an analogy from the fintech concept, insurtech can be defined as the integration of emerging technologies to enhance insurance services through targeted solutions addressing distinct problems within the insurance sector (Cao et al., 2020). Our future work will focus on developing threat models that contain the identified security threats and vulnerabilities in chatbots and mitigation strategies, culminating in formulating security requirements. In addition, we will conduct a security risk analysis of insurance chatbots using the STRIDE/DREAD combo. Traditional underwriting processes are often time-consuming and reliant on manual data collection and analysis. AI-driven data analytics streamlines these processes by automating data gathering, analysis, and decision-making.

Conversational search uses generative AI to free up non-technical business users from having to write and maintain answers manually, accelerating time to value and decreasing the total cost of ownership of virtual assistants. By automating routine tasks such as policy renewals, claims processing, and customer inquiries, insurers can reduce operational costs and improve efficiency. According to a report by Accenture, AI-driven automation could save the insurance industry up to $300 billion annually by 2030. Nuance Communications, known for their voice recognition healthcare NLP software called Nina, also offers a customer service virtual assistant. The assistant can be accessed via voice or text, and Nuance claims it can be integrated into a company’s website, smartphone app, text messaging application, or smart TV.

Most insurance companies have prioritized digital transformation and IT core modernization, using hybrid cloud and multi-cloud infrastructure and platforms to achieve the above-mentioned objectives . This approach can accelerate speed-to-market by providing enhanced capabilities for developing innovative products and services, facilitating business growth and improving the overall customer experience in their interactions with the company. The platform also has the capability to update users on the status of their loan or credit card application. A chatbot made with this platform can also answer more simple customer service questions such as locating a routing number. This application also covers refunds and can help the user check the status of a refund or request a new one. Users can also check the processing time of a given payment or refund, which may put their mind at ease while waiting for it to clear.

Then, the app compares the patient’s symptoms to their health profile as well as previous patient data regarding similar symptoms. Ada Health’s machine learning algorithms are trained on a database of thousands of medical symptoms and ailments, and expands that information over time using responses collected from each customer. We compiled a list of 20 chatbot applications currently being used across multiple industries. We explain what the chatbot software vendors offer their client companies and how that translates to customer satisfaction. This also includes internal chatbots made to help customer-facing employees work with their clientele.

Most developers do not make building security into the software right from the beginning a top priority46,47. A good reason for this is the lack of security experts or security requirements engineers in many organisations. Also, few requirements engineers/analysts have expertise in the areas of identifying, analysing, specifying, and managing security requirements47,48. Chatbots are mostly accessible through different platforms of messenger apps such as Facebook and Skype, and there is no proper security implementation on these platforms.

So far in the literature, none of the previous studies on the security of chatbots have considered using STRIDE for threat elicitation or focussed on the insurance industry. Thus, more understanding of insurance chatbots’ security threats and vulnerabilities is needed. We need to ensure that our self-service tools, such as chatbots and helpdesk articles, are actually solving our customers’ problems. And we need to do it in a way that is effective, efficient and convenient for our customers. Many large insurers are finding ways to digitize parts of their business process in preparation for future projects involving machine learning. This is especially true in claims processing, which could become faster and less error-prone if claims adjusters did not have to search through large amounts of data or paper documents manually.

The dire shortage of mental health professionals—particularly those willing to work with imprisoned people and those experiencing homelessness—is exacerbated by the amount of time providers need to spend on paperwork, Insel says. Programs such as ChatGPT could easily summarize patients’ sessions, write necessary reports, and allow therapists and psychiatrists to spend more time treating people. “We could enlarge our workforce by 40 percent by off-loading documentation and reporting to machines,” he says. The concept of “robot therapists” has been around since at least 1990, when computer programs began offering psychological interventions that walk users through scripted procedures such as cognitive-behavioral therapy. More recently, popular apps such as those offered by Woebot Health and Wysa have adopted more advanced AI algorithms that can converse with users about their concerns. And chatbots are already being used to screen patients by administering standard questionnaires.

By combining extensive data with sensory analytics and adaptive assessments, Lapetus Solutions empowers insurance companies to provide more accurate policies that are tailored to the unique needs of each individual. Bold Penguin allows insurance companies to quickly write policies that stand out in the industry with two AI-powered tools, SubmissionLink and ClauseLink. When carriers receive documents from agencies, SubmissionLink analyzes these materials and locates essential data points for underwriters. Meanwhile, ClauseLink reviews insurance clauses to help providers compare their policies to those of competitors.

A chatbot that spoke to them using natural language assisted them throughout the process, making it an easy and efficient experience. The same can be said of chatbots that help customers with claims or policy changes, though the technology can quickly become more personal when necessary. The benefits of implementing AI-powered chatbots for customers and their insurance companies are manifold. Allstate’s business insurance chatbot was developed in collaboration with Jorsek for customer service to a base of small business owners. Allstate customers can access ABIE through their website, and ask insurance-related questions that a live agent would normally answer. But their solutions are also targeted to other verticals, such as banking, telecom and insurance.

According to a report by PwC, blockchain technology can reduce fraud in the insurance industry by up to 30%, highlighting its potential to enhance trust and integrity. Smart contracts are self-executing contracts with the terms of the agreement directly written into code. These contracts run on blockchain networks, ensuring that all parties have access to the same information and that transactions are secure and transparent. In the insurance industry, smart contracts can automate various processes, such as policy issuance, claims processing, and premium payments. For example, Zurich Insurance has implemented a low-code platform to develop customer-facing applications and streamline internal processes. However, the benefits of low-code platforms extend beyond speed and efficiency; they also facilitate collaboration between IT and business teams, enabling them to work together to develop and deploy solutions that meet specific business needs.

The patients who were lying down were much more likely to be seriously ill, so the algorithm learned to identify COVID risk based on the position of the person in the scan. “Therapy is best when there’s a deep connection, but that’s often not what happens for many people, and it’s hard to get high-quality care,” he says. It would be nearly impossible to train enough therapists to meet the demand, and partnerships between professionals and carefully developed chatbots could ease the burden immensely. “Getting an army of people empowered with these tools is the way out of this,” Insel says.

With Sprout’s Bot Builder, you can enhance your customer care strategy and improve response times on important social channels like Facebook and X. Start your free 30-day trial today and see how chatbots can transform your customer service experience. Customers get speedy, efficient support for their common issues and agents get to focus on complex tasks only they can handle, increasing satisfaction for both parties. Customer-to-chatbot interactions will stream directly into Sprout’s Smart Inbox, supporting seamless handoff between bot and human support. If you’re using Sprout’s integration with Salesforce, you can gain a 360-degree understanding of specific customer experiences in just a few clicks.

Leading Insurers Are Having a Generative AI Moment – BCG

Leading Insurers Are Having a Generative AI Moment.

Posted: Thu, 17 Aug 2023 07:00:00 GMT [source]

After all, more and more people are looking for natural language interfaces when it comes to AI technology, says McGloin. Just take a look at Alexa and the types of conversations consumers are having with the device. Despite the massive venture investments going into healthcare AI applications, there’s little evidence of hospitals using machine learning in real-world applications. We decided that this topic is worth covering in depth since any changes to the healthcare system directly impact business leaders in multiple facets such as employee insurance coverage or hospital administration policies.

Regarding the topic of data security in chatbots, the study also makes an empirical contribution by offering new evidence from the geographical context of the South African insurance industry, which is not common in the literature. Practically, it also provides a good guide for designers and developers (particularly insurance chatbots) on critical aspects where financial chatbots can be vulnerable and susceptible to security attacks. In summary, we observe that none of the previous studies have focused on threat modelling for insurance chatbots. Also, no study has used STRIDE modelling to identify security threats that pertain to insurance chatbots. STRIDE, as the oldest and most mature threat modelling method25, has the capabilities to afford reliable, proactive security assessment of insurance chatbots. Thus, as a contribution, this paper presents a first attempt at threat modelling for insurance chatbots using STRIDE.

Additionally, the onboarding assistant helps potential customers set up their account with the client insurer and then likely directs them to the individual policies. Users can ask the chatbot questions about the details of each policy and possibly the type of customer that usually needs that policy. This allows the user to make these appointments and even purchase insurance without the need of a human employee, which saves time for the client insurer. This is likely convenient for Allstate in that new insurance information can be updated across the entire business using a single platform. This also likely allows ABIE to retrieve all information necessary to provide answers from one location.

Depending on the client’s needs, Passage’s platform can help build a chatbot that accurately sources and translates the right data into a fulfilling answer to the user. The titular Cleo chatbot can answer basic financial questions such as “can I afford a coffee? ” and it will respond with immediate calculations based on your incoming bills and account balance. In order to create an effective chatbot that made use of all relevant aspects of Watson, the company can include the Tone Analyzer tool for sentiment analysis, and run the entire application from the IBM cloud platform.