KeyCorp (KEY) Risk Analysis

We use quantitative models to help manage certain aspects of our business and to assist with certain business decisions, including, but not limited to, estimating ALLL, measuring the fair value of financial instruments when reliable market prices are unavailable, estimating the effects of changing interest rates and other market measures on our financial condition and results of operations, managing risk, predicting cash flows, and for capital planning purposes (including during the capital stress testing process). Models are simplified representations of real-world relationships. Thus, our modeling methodologies rely on many assumptions, historical analyses, correlations, and available data. These assumptions provide only reasonable, not absolute, estimates, particularly in times of market distress when historical correlations on which we rely may no longer be relevant. Additionally, as businesses and markets evolve, our measurements may not accurately reflect this evolution. Models can also produce inadequate estimates due to errors in computer code, use of unsuitable data during model development or implementation, or the use of a model for a purpose outside the scope of the model's design. Some models we use employ methodologies based on artificial intelligence or machine learning. Compared to traditional models, these models may involve some additional complexities, such as the need for large datasets for training, the potential for algorithmic bias, and difficulty in interpreting model outputs. If our models fail to produce reliable results on an ongoing basis, we may not make appropriate risk management, capital planning, or other business or financial decisions. Furthermore, strategies that we employ to manage and govern the risks associated with our use of models may not be effective or fully reliable, and as a result, we may realize losses or other lapses. We have an enterprise-wide model risk management program designed to accurately identify, measure, report, monitor, and manage model risk. The management of model risk includes independent validation and model governance, establishing and monitoring model control standards and model risk metrics, and completeness and accuracy of the inventory of models. Banking regulators continue to focus on the models used by banks and bank holding companies in their businesses. The failure or inadequacy of a model may result in increased regulatory scrutiny on us or may result in an enforcement action or proceeding against us by one of our regulators.

We regularly review and update our internal controls, disclosure controls and procedures, compliance monitoring activities, and corporate governance policies and procedures. We also maintain an ERM program designed to identify, measure, monitor, report, and analyze our risks. Additionally, our internal audit function provides an independent assessment and testing of Key's internal controls, policies, and procedures. Any system of controls and any system to reduce risk exposure, however well designed, operated, and tested, is based in part on certain assumptions and can provide only reasonable, not absolute, assurances that the objectives of the system are met. The systems may not work as intended or be circumvented by employees, third parties, or others outside of Key. Additionally, instruments, systems, and strategies used to mitigate or otherwise manage exposure to various types of risk could be less effective than anticipated. As a result, we may not be able to effectively or fully mitigate our risk exposures in particular market environments or against particular types of risk.Our operations and financial performance could be adversely affected by severe weather and natural disasters, both directly and as a result of impacts on our customers. Natural disasters, including wildfires, tornadoes, severe storms, and hurricanes, have seemingly become more frequent and severe. The timing and effects of these climate-related physical risks are difficult to accurately predict, and the potential impact of such risks on our operations, employees, communities, and customers could have a material adverse effect on our business, financial position, and results of operations. Given our broad regional focus, we are exposed to a wide range of climate-related physical risks across different geographical areas. Severe weather events can directly affect our operations by interrupting systems, damaging facilities, disrupting our supply chain, and hindering our ability to conduct business as usual. Additionally, these events can indirectly impact us by damaging or destroying customer businesses, impairing their ability to repay loans, or causing damage to properties pledged as collateral for loans made by Key. Although preventative measures may help to mitigate damage, such measures could be costly, and any disaster could adversely affect our ability to conduct our business as usual. Furthermore, the insurance we maintain may not be adequate to cover our losses resulting from any business interruption resulting from a natural disaster or other severe weather events. Recurring extreme weather events could also reduce or eliminate the availability or increase the cost of insurance to Key and our customers. Our failure to comply with evolving regulatory requirements related to natural disaster risk management may also result in legal and financial consequences.Our development and use of AI, including through third parties, exposes us to inherent risks that may adversely impact KeyCorp. We use, and will increasingly use AI, including through third party vendors acting on our behalf and other counterparties, in connection with our business and operations. AI is complex and rapidly evolving and in order to compete with other banks and financial institutions effectively, we must incorporate new and emerging AI technology into our business and this may subject us to new or heightened legal, regulatory, operational, and other risk. The legal and regulatory environment relating to AI is uncertain and evolving, and any changes to applicable laws and regulations could require changes to our use of AI technology and could cause an increase in associated costs and expenses. We may also be unsuccessful in realizing the intended benefits of AI or otherwise enhancing our business or operations and our competitors may incorporate AI in their businesses or operations more quickly or more successfully than us, all of which could occur despite considerable expense and which could negatively affect our financial condition and results of operations. The models underlying AI that we may leverage, including those developed by third party providers, may be incorrectly or inadequately designed or implemented and trained on, or otherwise use, data or algorithms that are incomplete, inadequate, misleading, biased, or otherwise flawed, or that are subject to intellectual property rights not known to us, and that ultimately produce outputs that are similarly flawed but that which we or third parties acting on our behalf rely, and any such flaw may not be easily and readily detectable. The limited transparency into the underlying complexity of AI and associated models that we rely on makes reproducing the connection between input and output, at times, difficult or impossible. If the AI that we leverage is flawed in such ways, we may make inaccurate or ineffective decisions and otherwise incur operational inefficiencies, compliance issues, competitive and reputational harm, adverse legal and regulatory actions, or other adverse impacts to our business and operations. We may not be able to sufficiently mitigate or detect any of the foregoing risks given the emerging nature of AI technology. Additionally, inappropriate or controversial data practices by third party AI developers and their end-users could adversely affect public opinion of AI and ultimately impair acceptance of AI, including those incorporated into our business and operations.