Biometric scoring: Behavioral biometrics might verify identities more accurately

IMAGE CREDIT:
Image credit
iStock

Biometric scoring: Behavioral biometrics might verify identities more accurately

Biometric scoring: Behavioral biometrics might verify identities more accurately

Subheading text
Behavioral biometrics such as gait and posture are being studied to see if these non-physical characteristics can improve identification.
    • Author:
    • Author name
      Quantumrun Foresight
    • February 13, 2023

    Insight summary

    Behavioral biometric data may reveal patterns in people’s actions and reveal a lot about who they are, what they are thinking, and what they will likely do next. Behavioral biometrics employ machine learning to interpret hundreds of distinct biometric measurements to identify, authenticate, nudge, reward, and punish.

    Biometric scoring context

    Behavioral biometric data is a technique for analyzing even the smallest variations in human behavior. The phrase is frequently contrasted to physical or physiological biometrics, which describes human features like iris or fingerprints. Behavioral biometrics tools can identify individuals based on patterns in their activity, such as gait or keystroke dynamics. These tools are increasingly used by financial institutions, businesses, governments, and retailers for user authentication. 

    Unlike traditional verification technologies that work when a person's data is collected (e.g., pressing a button), behavioral biometric systems can authenticate automatically. These biometrics compare an individual’s unique pattern of behavior to past behavior to establish their identity. This process can be done continuously throughout an active session or by recording specific behaviors.

    The behavior may be captured by an existing device, like a smartphone or laptop, or by a dedicated machine, such as a sensor designed specifically for measuring footfalls (e.g., gait recognition). The biometric analysis produces a result that reflects the likelihood that the individual performing the actions is the one who established the system's baseline behavior. If a customer's behavior falls outside of the expected profile, additional authentication measures will be put into place, such as fingerprint or facial scans. This feature would better prevent account takeover, social-engineering scams, and money laundering than traditional biometrics.

    Disruptive impact

    A behavior-based approach, such as movements, keystrokes, and phone swipes, can help authorities identify someone securely in situations where physical characteristics are hidden (e.g., use of face masks or gloves). In addition, solutions that rely on keystrokes for computer-based identity verification have shown to be able to identify individuals based on their typing habits (the frequency and rhythms seem unique enough to establish identification). Because typing is a form of data input, the algorithms can improve as they continue to track and analyze keystroke information.

    However, in certain instances, the context restricts the accuracy of this behavioral biometric. Individual patterns on different keyboards may vary; physical conditions such as carpal tunnel syndrome or arthritis may impact movement. It is tough to compare the various providers' trained algorithms without standards.

    Meanwhile, image recognition provides analysts with greater amounts of data that can be used for behavioral research. Even though they aren't as accurate or reliable as other biometric approaches, gait and posture biometrics are becoming increasingly useful tools. For example, these features can be enough to establish identity in crowds or public places. Police forces in countries that implement the European Union (EU)’s General Data Protection Regulation (GDPR) use biometric data, such as gait and movement, to immediately assess threatening situations.

    Implications of biometric scoring

    Wider implications of biometric scoring may include: 

    • Increasing concerns about artificial intelligence (AI)’s potential to misidentify/misunderstand human behavior, especially in law enforcement, which can lead to wrongful arrests.
    • Fraudsters mimicking gait and keyboard typing rhythms to infiltrate systems, particularly in financial institutions.  
    • Biometric scoring expanding into consumer scoring where people with disabilities/limited mobility can be discriminated against.
    • Increasing debates on whether behavioral biometric data, including heart rates, can be included in digital privacy regulations.
    • People being able to log into websites and apps just by typing in their usernames.

    Questions to consider

    • Do you agree that behavioral biometrics will be more useful for identity verification?
    • What other potential problems can this type of biometric identification have?

    Insight references

    The following popular and institutional links were referenced for this insight: