Distracted driving—which has significantly increased since the coronavirus pandemic—is most significantly affected by cellphone use, according to a new Issues Brief by Triple-I.
The report, Distracted Driving: State of the Risk, states that cellphone use–which includes dialing, texting, and browsing–was among the most ubiquitous and highest-risk behaviors found in governmental and private sector studies. According to a 2022 national observational survey from the National Highway Traffic Safety Administration (NHTSA), a total of 2.5 percent of drivers stopped at intersections were talking on hand-held phones at any moment during the day in 2021.
The brief also found that the U.S. personal auto insurance industry’s combined ratio—a measure that represents underwriting profitability—increased dramatically from 2022, to 112.2. A combined ratio below 100 indicates an underwriting profit, while one above 100 indicates an underwriting loss.
“As drivers returned to the roads following the pandemic, distracted driving surged, causing higher rates of accidents, injuries, and deaths. This high-risk behavior has worsened in the years since, having huge implications for the insurance industry and their policyholders,” stated Dale Porfilio, chief insurance officer, Triple-I.
The report notes that telematics and usage-based insurance can potentially help insurers—and their policyholders—better understand a driver’s risk profile and tailor auto insurance rates based on individual driving habits.
Indeed, according to an Insurance Research Council survey in 2022, 45 percent of drivers said they made significant safety-related changes in how they drove after participating in a telematics program. An additional 35 percent stated that they made small changes in their driving behavior. Policyholders became more comfortable with having their insurer monitor their driving behavior when it resulted in potentially lower insurance costs during the onset of the pandemic.
“If telematics can influence drivers to change behaviors and reduce the number of accidents, the nation’s roadways will be safer and auto insurance can be more affordable,” Porfilio concluded.
The insurance industry’s shift from assessing and pricing risks to predicting and preventing losses – thereby improving insurance availability and affordability – is well underway. Even a casual look at the trade press reveals insurers adopting technologies and data-driven strategies that help businesses, families, and communities improve their risk profiles.
This data-driven movement does more than simply contain insurance costs – it’s driving improved customer engagement, affinity, and retention and creating opportunities beyond the transactional. Data clarity is crucial for all stakeholders, from insurers to first responders utilities, policymakers and – most important – homeowners. Accurate data enables proactive measures that can prevent fires from happening.
We’re seeing this with our insurance IoT offering, Ting. Ting prevents home fires by identifying unique signals generated by tiny electrical arcs, the precursors to imminent fire risks. These signals are incredibly small but are clearly visible to Ting’s advanced detection technology. Ting has been found to prevent 80 percent of home electrical fires – and, beyond its ability to predict and prevent, we have found that Ting holds even greater significance for organizations that want to bring greater clarity and value to their current data ecosystems.
Over the past few years, we’ve built the world’s most knowledgeable electrical fire prevention team, which has been instrumental in the evolution of Ting’s machine learning and AI. Our Fire Safety Team has found that existing electrical fire data, while helpful and directional, needs greater accuracy and completeness. This is not due to a lack of care. We’re talking about an exceptionally hard problem – codifying fires after the fact. It is at this critical point where data from IoT devices like Ting becomes indispensable.
More than 50 percent of insurance claims for fire are often coded in the “unknown/underdetermined” category. Of these, fire chiefs and forensic fire engineers suggest more than half are likely electrical-related, but lack of resources prevent them from determining exact causation beyond a reasonable doubt, so they simply default to “unknown.” Ting data continues to document important and first of its kind findings around the origin of electrical fires.
Our ‘why’ behind predict and prevent
A horrific loss from an electrical fire in my family prompted the question: “Why can’t faults be identified well before they can evolve into a fire?”
Electricity is one of the most dangerous forces in nature, yet one of our most critical resources; our growing reliance poses increasing risks to homes, businesses, and communities. Recent U.S. Fire Administration data reveals a sobering trend. The 10 years from 2012 through 2021 saw reduced cooking, smoking, and heating fires; however, in stark contrast, electrical fires saw an 11 percent increase over that same period. Fire ignitions with an undetermined cause increased equally by 11 percent.
Our pursuit to address these trends has brought us and our insurance partners here: Nearly 400,000 home-years of data, 6,000 remediated hazards; an insurance-forward IoT and telematics platform with full turnkey delivery; and most notably, hundreds of thousands of customers thrilled that their insurance company is doing more for them than reactively paying claims.
Beyond the home’s walls
But Ting’s value is not limited to inside the home. While every Ting sensor is monitoring each home’s electrical activity to help predict and prevent fires, collectively the Ting network is aggregating data from across the broader utility grid. Specifically, it can help predict and prevent faults on the grid, enabling operators to proactively address risks that might otherwise lead to catastrophic, loss-generating events like wildfires.
Data drives insights
Given that many electrical-related fires are coded not as electrical but as “unknown” in fire incident databases, we’ve learned that comparing “prevented fires” to claims after a fire must consider a broader set of fire claims across a book of business, not just those with a secondary cause of “electrical.” All unknown fires and any claim that could even be electrical-related should be included in the broader set of claims. Excluding claims that can reasonably and accurately be removed — such as arson, lightning, earthquakes, and wildfire-related home fires — the data reveals a one-third reduction in the broader category of fires across the “Ting cohort” versus the “non-Ting cohort.” This results in a strong ROI for insurers.
Beyond prevention metrics, we’ve learned a lot, and Ting continues to learn daily and provide statistically significant actuarial impacts. With fully documented and mitigated hazards identified in 1 in 68 homes, the cases – or “saves” – are documented in detail in a peer-reviewed whitepaper, the latest version published on June 1, 2023. By design, each identified and remediated hazard is carefully reported through a highly standardized process to ensure high-quality, consistent data.
Upon analyzing this statistically significant data, a recurring theme surfaced: The longstanding perception of the electrical fire problem requires new thinking. Below, I highlight three surprising, objective observations revealed by Ting data that support this notion:
There is a common misconception that electrical fires are largely due to older home wiring infrastructure. Yet, we have found that 50 percent of home electrical fire hazards stem from failing or defective devices and appliances, with the other half attributed to home wiring and outlets. This finding is reflected in the chart below, breaking down the location and types of home electrical fire hazards, with a breakout of those stemming from devices and appliances.
What may seem more surprising is that the electric utility grid can be a significant fire risk factor inside the home – not just a community fire risk. Nearly 50 percent of all hazard cases trace back to a root cause outside the house in the form of a grid equipment fault. These faults result in dangerous power entering the home. These conditions endanger a home and its occupants and can cause a shock hazard, damage equipment, and sensitive electronics, and worse, ignite a fire. Utility repair crews often share that a hazard impacted multiple homes in the immediate area, not just the home protected by Ting.
One last finding that runs counter to conventional thinking about electrical fire risk comes in the form of a home-age “bias.” Logically, most of us assume the older the home, the higher the risk. In general, this holds when considering the effects of age and use on existing wiring infrastructure – all other things being equal. However, this assumption falls apart when considering all other factors, such as materials, build quality, and the standards and codes at that time. In fact, with the prevention data that flows in each day from our Fire Safety Team, we have built predictive models for home fire risk; early indications are that these models are demonstrating skill and will lead to a better, more informed view of risk – and of course – even better prevention.
I’m amazed at how our initial objective preventing residential fires has evolved to take on such a broad scope. New data spawns new thinking and new opportunities. Objective data is essential to validating the efficacy of any initiative seeking to prevent losses. Predicting and preventing fires is in the interest of all – especially homeowners and their families.
The Colorado Division of Insurance’s recent adoption of regulations to govern life insurers’ use of any external consumer data and information sources is the first step in implementing legislation approved in 2021 aimed at protecting consumers in the state from insurance practices that might result in unfair discrimination.
Property/casualty insurers doing business in Colorado should be keeping an eye on how the legislation is implemented, as rules governing their use of third-party data will certainly follow.
The implementation regulations, which have been characterized as a “scaling back” of a prior draft release in February, require life insurers using external data to establish a risk-based governance and risk-management framework to determine whether such use might result in unfair discrimination with respect to race and remediate unfair discrimination, if detected. If the insurer uses third-party vendors and other external resources, it is responsible under the new rules for ensuring all requirements are met.
Life insurers must test their algorithms and models to evaluate whether any unfair discrimination results and implement controls and process to adjust their use of AI, as necessary. They also must maintain documentation including descriptions and explanations of how external data is being used and how they are testing their use of external data for unfair discrimination. The documentation must be available upon the regulator’s request, and each insurer must report its progress toward compliance to the Division of Insurance.
The revised draft no longer focuses on “disproportionately negative outcomes” that would have included results or effects that “have a detrimental impact on a group” of protected characteristics “even after accounting for factors that define similarly situated consumers.” Removing that term altogether, the revised draft shifts focus to requiring “risk-based” governance and management frameworks.
This change is significant. As Triple-I has expressed elsewhere, risk-based pricing of insurance is a fundamental concept that might seem intuitively obvious when described – yet misunderstandings about it regularly sow confusion. Simply put, it means offering different prices for the same level of coverage, based on risk factors specific to the insured person or property. If policies were not priced this way – if insurers had to come up with a one-size-fits-all price for auto coverage that didn’t consider vehicle type and use, where and how much the car will be driven, and so forth – lower-risk drivers would subsidize riskier ones.
Risk-based pricing allows insurers to offer the lowest possible premiums to policyholders with the most favorable risk factors. Charging higher premiums to insure higher-risk policyholders enables insurers to underwrite a wider range of coverages, thus improving both availability and affordability of insurance. This straightforward concept becomes complicated when actuarially sound rating factors intersect with other attributes in ways that can be perceived as unfairly discriminatory.
Algorithms and machine learning hold great promise for ensuring equitable pricing, but research has shown these tools also can amplify any biases in the underlying data. The insurance and actuarial professions have been researching and attempting to address these concerns for some time (see list below).
Want to know more about the risk crisis and how insurers are working to address it? Check out Triple-I’s upcoming Town Hall, “Attacking the Risk Crisis,” which will be held Nov. 30 in Washington, D.C.
Artificial intelligence is helping to limit the costs associated with data breaches, a recent study by IBM and the Ponemon Institute found. While these costs continue to rise, they are increasing more slowly for some organizations – in particular, those using less-complex, more-automated security systems.
According to the study, the average cost of a data breach was $4.45 million in 2023, a 2.3 percent increase from the 2022 cost of $4.35 million. The 2023 figure represents a 15.3 percent increase from 2020, when the average breach was $3.86 million.
However, not all organizations surveyed by the study experienced the same kinds of breaches – or the same costs. Organizations with “low or no security system complexity” – systems in which it is easier to identify and manage threats – experienced far smaller losses than those with high system complexity. The average 2023 breach cost $3.84 million for the former and a staggering $5.28 million for the latter. For organizations with high system complexity, this is an increase of more than 31 percent from the year before, amounting to an average of $1.44 million.
As David W. Viel, founder and CEO of Cognoscenti Systems, put it: “The size and complexity of a system directly results in a greater number of defects and resulting vulnerabilities as these quantities grow. On the other hand, the number of defects and cybersecurity vulnerabilities shrinks as the system or component is made smaller and simpler. This strongly suggests that designs and implementations that are small and simple should be very much favored over large and complex if effective cybersecurity is to be obtained.”
The research also noted that organizations that involve law enforcement in ransomware attacks experienced lower costs. The 37 percent of survey respondents that did not contact law enforcement paid 9.6 percent more than those that did, with the breach lasting an average of 33 days longer than those that did contact law enforcement. These longer breaches tended to cost organizations far more, with breaches with identification and containment times under 200 days averaging $3.93 million, and those over 200 days costing $4.95 million.
AI and automation are proving key
Security AI and automation both showed to be significant factors in lowering costs and reducing time to identify and contain breaches, with organizations utilizing these tools reporting 108-day shorter times to contain the breach, and $1.76 million lower data breach costs relative to organizations that did not use these tools. Organizations with no use of security AI and automation experienced an average of $5.36 million in data breach costs, 18.6 percent more than the average 2023 cost of a data breach.
Now, most respondents are using some level of these tools, with a full 61 percent using AI and automation. However, only 28 percent of respondents extensively used these tools in their cybersecurity processes, and 33 percent had limited use. The study noted that this means almost 40 percent of respondents rely only on manual inputs in their security operations.
Cyber insurance demand is growing
A recent study by global insurance brokerage Gallagher showed that the vast majority of business owners in U.S. – 74 percent – expressed extreme or very high concern about the impact of cyberattacks on their businesses. Indeed, a study by MarketsandMarkets found that the cyber insurance market is projected to grow from $10.3 billion in 2023 to $17.6 billion by 2028, noting that the rise in threats like data breaches, ransomware, and phishing attacks is driving demand.
Organizations are now responding more thoroughly to these threats, with increased underwriting rigor helping clients progress in cyber maturity, according to Aon’s 2023 Cyber Resilience Report. Aon states that several cybersecurity factors, including data security, application security, remote work, access control, and endpoint and systems security – all of which experienced the greatest improvement among Aon’s clients – must be continually monitored and evaluated, particularly for evolving threats.
Insurers and their customers need to work together to more fully address the risks and damages associated with cyberattacks as these threats continue to grow and businesses rely ever more heavily on technology.
Insurance agencies that adopt digital methods to interact with customers have seen their revenues grow faster than their less digitally sophisticated competitors, according to new research by Liberty Mutual and Safeco Insurance. However, the research also indicates that digital adoption by agencies has slowed in recent years.
The study, The State of Digital in Independent Insurance Agencies, found that “highly digital adopter” agencies — based on a 10-point scale related to the number and complexity of the tools the agency uses — experienced a 70 percent growth rate, as opposed to 17 percent for “high digital adopters”, and a mere 10 percent for “low” and “medium” digital adopters.
But while digital adoption has gained traction, it has declined as a priority in agencies’ plans. In the latter part of 2020, 58 percent of agencies said improving digital capabilities was part of their five-year growth plans, according to the Liberty Mutual/Safeco study. However, by late 2021, this had decreased to 47 percent, approximately the same as in 2017.
The digital tools that have seen a decrease in use range from social media to live online chats. Additionally, many agencies said they are not tracking which digital tools are driving growth.
The survey found that 60 percent of digitally focused agencies said they planned to invest in new digital capabilities within their five-year agency growth plans. Only 42 percent of slow and steady growth agencies said the same. Growth-focused agencies have used several tools to increase their reach and revenue. Self-service portals, video calls, live online chats, video quotes, and policy reviews have all driven significant improvement among these agencies.
These, however, are not the only tools being recommended and used. Artificial intelligence, machine learning, Internet of Things, and big data analytics are all being considered and used to increase engagement with customers and prospects.
Cybercrime may be a factor hampering growth in digital adoption. Indeed, global cybercrime costs are predicted to hit $10.5 trillion annually by 2025, according to Cybersecurity Ventures. Additionally, more than half of all consumers have experienced a cybercrime at some point, according to a 2021 survey by Norton.
Agents remain alert to cyber threats. The Liberty Mutual/Safeco study found that 57 percent of survey respondents anticipated that cyber liability would have a major impact on their agencies by 2025, an increase from 46 percent in 2017.
As more new vehicles become equipped with crash-avoidance features, some owners report significant issues with the technologies after repairs, according to a recent report from the Insurance Institute of Highway Safety (IIHS).
In the survey, approximately half of those who reported an issue with equipped front crash prevention, blind-spot detection, or rearview or other visibility-enhancing cameras said at least one of those systems presented problems after the repair job was completed.
Nevertheless, many owners remained eager to have a vehicle with these features and were pleased with the out-of-pocket cost, according to Alexandra Mueller, IIHS senior research scientist.
“These technologies have been proven to reduce crashes and related injuries,” Mueller said. “Our goal is that they continue to deliver those benefits after repairs and for owners to be confident that they’re working properly.”
Still, as problems with these technologies persist, the study notes that it is important to track repair issues to further the adoption of crash avoidance features. IIHS research has shown that front-crash prevention, blind-spot detection, and rearview cameras all substantially reduce the types of crashes they are designed to address. For example, IIHS said, automatic emergency braking reduces police-reported rear-end crashes by 50 percent.
An analysis conducted by the IIHS-affiliated Highway Loss Data Institute (HLDI) showed the reduction in insurance claims associated with Subaru and Honda crash-avoidance systems remained essentially constant, even in vehicles more than five years old. But repairs can make it necessary to calibrate the cameras and sensors that the features rely on to work properly, making repairs complicated and costly.
For example, a simple windshield replacement can cost as little as $250, while a separate HLDI study found vehicles equipped with front crash prevention were much more likely to have glass claims of $1,000 or more. Much of that higher cost is likely related to calibration.
The new IIHS study found that owners often had more than one reason requiring repairs to these safety features. Most had received a vehicle recall or service bulletin about their feature, but that was rarely the sole reason they brought their vehicles in for service or repair.
“Other common reasons — which were not mutually exclusive — included windshield replacement, crash damage, a recommendation from the dealership or repair shop, and a warning light or error message from the vehicle itself,” according to the study.
Repair difficulties could motivate drivers to turn off crash avoidance features, potentially making collisions more likely. But, despite the post-repair issues, the study found that slightly more than 5 percent of owners would opt not to purchase another vehicle with the repaired feature. As reckless driving and traffic fatalities continue to rise, advanced driver-assistance systems will only become more important for the roadway safety, necessitating reliable technology.
Too many car owners are too comfortable leaving their vehicles’ driver-assist features in charge, potentially putting themselves and others at risk, according to the Insurance Institute for Highway Safety (IIHS).
IIHS said a survey of about 600 regular users of General Motors Super Cruise, Nissan/Infiniti ProPILOT Assist, and Tesla Autopilot found they were “more likely to perform non-driving-related activities like eating or texting while using their partial automation systems than while driving unassisted.”
“The big-picture message here is that the early adopters of these systems still have a poor understanding of the technology’s limits,” said IIHS President David Harkey.
The study reports that 53 percent of Super Cruise users, 42 percent of Tesla Autopilot users, and 12 percent of Nissan’s ProPilot Assist users were comfortable letting the system drive without watching what was happening on the road. Some even described being comfortable letting the vehicle drive during inclement weather.
These systems combine adaptive cruise control and lane-keeping systems, primarily to keep a car in a lane and following traffic on the highway. All require an attentive human driver to monitor the road and take full control when called for.
“None of the current systems is designed to replace a human driver or to make it safe for a driver to perform other activities that take their focus away from the road,” IIHS said in announcing the results of its survey.
While all three automakers caution drivers about the systems’ limits, confusion remains. Tesla’s driver-assist system, which it calls “full self-driving” has received much scrutiny over the years as auto safety experts say the name is misleading and risks worsening road safety.
The U.S.government has set no standards for these features, which are some of the newest technologies on vehicles today. A patchwork of state laws and voluntary federal guidelines is attempting to cover the testing and eventual deployment of autonomous vehicles in the United States.
Insurance fraud costs the U.S. $308.6 billion a year, according to recent research by the Coalition Against Insurance Fraud (CAIF). And, while staffing within insurers’ Special Investigation Units (SIU) is a pain point, CAIF found that use of anti-fraud technology is on the rise.
CAIF notes that hardest-hit insurance lines are:
Life insurance, at $74.7 billion annually;
Medicare and Medicaid, at $68.7 billion; and
Property and casualty, $45 billion.
“There is a huge and monumental impact that insurance fraud causes to American citizens, American families, and to our economy every single year,” said Matthew Smith, the coalition’s executive director.
Another recent CAIF study looked at SIUs and insurers’ response to fraud. The study found that SIU staff grew at 1.4 percent from 2021 to 2022, slower than the 2.5 percent growth rates from two previous studies addressing this issue. Staffing and talent are among the top concerns of anti-fraud leaders CAIF surveyed.
However, an additional CAIF study found that anti-fraud technology is increasingly being used—a positive sign in the fight against these crimes. Among the key findings of that report is that 80 percent of respondents use predictive modeling to detect fraud, up from 55 percent in 2018.
Insurance fraud is not a victimless crime. According to the FBI, the average American family spends an extra $400 to $700 on premiums every year because of fraud. Most of these costs are derived from common frauds, including inflating actual claims; misrepresenting facts on an insurance application; submitting claims for injuries or damage that never occurred; and staging accidents.
To further combat insurance fraud, there are ways to file complaints, including contacting your state’s fraud bureau; contacting your insurer to see if a fraud system is in place; using the National Insurance Crime Bureau (NICB) “Report Fraud” button; and reporting it to a local FBI branch.
“Insurance fraud is the crime we all pay for,” CAIF’s Smith added. “Ultimately, it’s American policyholders and consumers that pay the high cost of insurance fraud.”
State prohibitions on cellphone use while driving correlate with reduced crash rates, according to recent research by the Insurance Institute for High Safety (IIHS). However, overall results were mixed among the states studied, with different legal language, degrees of enforcement, and penalty severity, providing possible explanations for the differing outcomes.
The study observed crash rate changes in California, Oregon, and Washington after legislation to prevent cellphone calls and texting while driving was enacted in 2017, with the research looking at overall numbers from 2015 to 2019. These numbers were compared to control states Idaho and Colorado.
Notably, the study found:
A 7.6 percent reduction in the rate of monthly rear-end crashes of all severities relative to the rates in the control states;
Law changes in Oregon and Washington were associated with significant reductions of 8.8 percent and 10.9 percent, respectively;
California did not experience changes in rear-end crash rates of all severities or with injuries associated with the strengthened law.
Still, state governments face several hurdles in their efforts to prevent crashes caused by cellphone use.
“Technology is moving much faster than the laws,” said Ian Reagan, a senior research scientist at IIHS. “Our findings suggest that other states could benefit from adopting broader laws against cellphone use while driving, but more research is needed to determine the combination of wording and penalties that is most effective.”
Distracted driving remains a major issue
Distracted driving remains a significant problem on roads nationwide. Indeed, distracted driving increased more than 30 percent from February 2020 to February 2022, due largely to changes in driving patterns spurred by the coronavirus pandemic, according to research by telematics service provider Cambridge Mobile Telematics.
The Governors Highway Safety Association (GHSA) reported that more than 3,100 people died in distraction-related accidents in 2020, with an estimated 400,000 people injured each year in such crashes. The true numbers, according to the study, are likely higher due to underreporting. The report also found that cell dial, cell text, and cell-browse were among the most prevalent and highest-risk behaviors.
Telematics can help
Telematics, which uses mobile technology to track driver behavior and provide financial incentives to drive less and often and more carefully, can help reduce dangerous driving. The more consumers positively react to the incentive, the less they pay for their insurance.
Research from the Insurance Research Council – like Triple-I, a nonprofit affiliate of The Institutes, focused on this exact issue, studying public perception and use of telematics. The study found that 45 percent of drivers surveyed said they made significant safety-related changes in the way they drove after participating in a telematics program. Another 35 percent said they made small changes in the way they drive.
During the pandemic, insurance consumers’ comfort with the idea of letting their driving be monitored in exchange for a better premium appeared to improve. In May 2019, mobility data and analytics firm Arity surveyed 875 licensed drivers over the age of 18 to find out how comfortable they would be having their premiums adjusted based on telematics variables. Between 30 and 40 percent said they would be either very or extremely comfortable sharing this data. In May 2020, they ran the survey again with more than 1,000 licensed drivers.
“This time,” Arity said, “about 50 percent of drivers were comfortable with having their insurance priced based on the number of miles they drive, where they drive, and what time of day they drive, as well as distracted driving and speeding.”
Some good news on the deepfake front: Computer scientists at the University of California have been able to detect manipulated facial expressions in deepfake videos with higher accuracy than current state-of-the-art methods.
Deepfakes are intricate forgeries of an image, video, or audio recording. They’ve existed for several years, and versions exist in social media apps, like Snapchat, which has face-changing filters. However, cybercriminals have begun to use them to impersonate celebrities and executives that create the potential for more damage from fraudulent claims and other forms of manipulation.
Deepfakes also have the dangerous potential to be used to in phishing attempts to manipulate employees to allow access to sensitive documents or passwords. As we previously reported, deepfakes present a real challenge for businesses, including insurers.
Are we prepared?
A recent study by Attestiv, which uses artificial intelligence and blockchain technology to detect and prevent fraud, surveyed U.S.-based business professionals concerning the risks to their businesses connected to synthetic or manipulated digital media. More than 80 percent of respondents recognized that deepfakes presented a threat to their organization, with the top three concerns being reputational threats, IT threats, and fraud threats.
Another study, conducted by a CyberCube, a cybersecurity and technology which specializes in insurance, found that the melding of domestic and business IT systems created by the pandemic, combined with the increasing use of online platforms, is making social engineering easier for criminals.
“As the availability of personal information increases online, criminals are investing in technology to exploit this trend,” said Darren Thomson, CyberCube’s head of cyber security strategy. “New and emerging social engineering techniques like deepfake video and audio will fundamentally change the cyber threat landscape and are becoming both technically feasible and economically viable for criminal organizations of all sizes.”
What insurers are doing
Deepfakes could facilitate the filing fraudulent claims, creation of counterfeit inspection reports, and possibly faking assets or the condition of assets that are not real. For example, a deepfake could conjure images of damage from a nearby hurricane or tornado or create a non-existent luxury watch that was insured and then lost. For an industry that already suffers from $80 billion in fraudulent claims, the threat looms large.
Insurers could use automated deepfake protection as a potential solution to protect against this novel mechanism for fraud. Yet, questions remain about how it can be applied into existing procedures for filing claims. Self-service driven insurance is particularly vulnerable to manipulated or fake media. Insurers also need to deliberate the possibility of deep fake technology to create large losses if these technologies were used to destabilize political systems or financial markets.
AI and rules-based models to identify deepfakes in all digital media remains a potential solution, as does digital authentication of photos or videos at the time of capture to “tamper-proof” the media at the point of capture, preventing the insured from uploading their own photos. Using a blockchain or unalterable ledger also might help.
As Michael Lewis, CEO at Claim Technology, states, “Running anti-virus on incoming attachments is non-negotiable. Shouldn’t the same apply to running counter-fraud checks on every image and document?”
The research results at UC Riverside may offer the beginnings of a solution, but as one Amit Roy-Chowdhury, one of the co-authors put it: “What makes the deepfake research area more challenging is the competition between the creation and detection and prevention of deepfakes which will become increasingly fierce in the future. With more advances in generative models, deepfakes will be easier to synthesize and harder to distinguish from real.”
