Experian's tech chief defends credit scores: 'We're not Palantir'

Summary of Experian's tech chief defends credit scores: 'We're not Palantir'

by The Verge

1h 9m•January 26, 2026
Listen to Episode

Overview of Decoder — "Experian's tech chief defends credit scores: 'We're not Palantir'"

This episode is an interview with Alex Lintner, CEO of Technology and Software Solutions at Experian, about how Experian collects and uses consumer data, its organizational structure, and how the company is approaching AI, privacy, security, and trust. The host probes the tension between the utility of credit data (access to credit, mortgages, car loans) and the power, opacity, and risks that come with centralized, large-scale data platforms.

Key topics covered

  • What Experian is today: a global data and technology company serving consumers and businesses across financial services, healthcare, automotive, and marketing.
  • Core product framing: data + analytics + AI as a platform to turn complex data into actionable guidance for lenders and consumers.
  • Company structure: a federated model with central functions (finance, HR, technology) and regional/business unit autonomy; technology executive board for governance.
  • AI use-cases and limitations: AI viewed as a platform capability (governance, explainability, oversight) rather than an autonomous decision-maker.
  • Consumer products: Experian Boost (free) that lets consumers add recurring payments to help improve credit scores; claims to be a real-time bureau.
  • Privacy/security posture: prioritizes privacy, consent, security; data sharding, encryption, cloud migration, and acquisition of anti-bot tech (NuraID).
  • Trust and accountability: Experian positions itself as providing information to decision-makers (lenders) not making the decisions, rejects “reputation score” comparisons (e.g., Palantir/Chinese social credit).

Main takeaways

  • Experian emphasizes that it is a data-and-platform company that supplies information and tools to lenders and consumers; it does not claim to make final decisions on behalf of lenders.
  • AI is used primarily to improve governance, detect model drift, provide explainability, and augment human oversight — not to fully automate credit decisions without human involvement.
  • Experian says it does not feed its data to public large language models and uses internal/smaller models and many task-specific agents (Lintner mentioned ~200 agents).
  • Top company priorities are privacy, consent, and security; security spending is the “first dollar” and considered an enabling cost for other products.
  • Experian portrays certain products (e.g., Boost) as consumer-empowering and free, designed to help underbanked or new-to-credit consumers build scores faster.
  • The company recognizes risks of scale (greater attack surface, increased power) and argues that scale is necessary for efficient credit markets that support prosperity.

How Experian uses AI (technical and governance aspects)

  • AI as platform capability: centralize model reuse, governance, and oversight rather than letting every region spin up uncontrolled AI.
  • Primary AI functions described:
    • Monitoring model performance and detecting model drift in real time.
    • Explaining which variables cause drift and prompting human review/adjustments.
    • Providing tooling for human data scientists to verify and gate models before production.
  • Model types: Lintner said Experian has internal LLMs and “small language models” (SLMs) for narrow tasks; emphasizes SLMs and agents for operational tasks rather than public LLMs for math-heavy calculations.
  • Testing and staging: use depersonalized and synthetic data for testing; nothing goes into production until data scientists validate it.

Data, privacy, and security practices

  • Core principles: keep privacy, consent, and security at the center of everything.
  • Technical measures discussed:
    • Encryption, data sharding (splitting bits of a profile across storage), strict access rights.
    • Cloud migration (AWS) to modernize infrastructure and security posture.
    • Purchase of anti-bot technology (NuraID).
  • Consumer controls: credit freeze, credit lock, opt-out options; Experian claims to make these accessible online and by phone, and runs U.S.-based call centers.
  • Security posture: security spend prioritized and defended internally; Experian positions itself as experienced in breach response and identity protection services.

Company structure and decision-making

  • Federated model: central standards and governance (finance, HR, tech) + regionally tailored business units that build customer-facing products.
  • Scale: ~23,000 employees; ~11,000 in technology organizations (4,000 direct to Lintner’s group + 7,000 in business-unit tech teams).
  • Governance forums: monthly Technology Executive Board chaired by Lintner with CTOs, CISO, risk officers, and responsible leads—used to align standards, roadmaps, and migrations.
  • Decision framework: principle-based leadership (privacy/consent/security highest), consensus-oriented, listen-heavy culture, and buy-in over autocratic edicts.

Notable quotes / memorable lines

  • "We're not Palantir." — Experian rejects being equated to a surveillance/reputation-score company.
  • "Keep privacy, consent and security at the center of everything we do." — Lintner’s repeated triad of priorities.
  • Origin anecdote: credit-scoring roots traced to a 19th-century merchant (C. Rommel) who recorded repayment behaviors — used to illustrate credit scoring as behavior-based, not identity-profile-based.
  • Experian’s stance: "We provide information. You provide the decision." — framing Experian as data provider to lenders, not decision-maker.

Where the risks are acknowledged

  • AI hallucination, bias, and mathematical weakness of LLMs: Lintner admits models (and LLMs) can be bad at certain tasks (math, increments) and must be human-supervised.
  • Scale increases risk and responsibility: larger data sets and integration increase attack surface and power concentration.
  • Migration and standardization are “spicy” and costly—technical migrations and decommissioning beloved tools are major internal conflicts.
  • Consumer perception: many people feel powerless or lack recourse when credit scores hurt them; Experian points to product fixes (Boost, real-time updates) and channels for resolution.

Practical recommendations / action items

  • For consumers:
    • Consider enrolling in Experian Boost (free) to include recurring payments that may improve scores.
    • Use credit freeze/lock if concerned about identity theft; reach out to customer support channels for disputes or corrections.
    • Regularly review your credit reports and use real-time features where available to accelerate fixes.
  • For technology/product builders:
    • Treat AI as a platform capability — centralize governance, reuse models where appropriate, and require human-in-loop oversight.
    • Invest in synthetic/depersonalized test data and robust QA before production deployments.
    • Prioritize encryption, access control, data sharding, and bot-detection defenses as foundational costs.
  • For policymakers and advocates:
    • Focus on transparency, recourse, and auditability of models used in lending decisions.
    • Ensure opt-out and redress mechanisms are easy and meaningful, especially for marginalized groups.

Final assessment

This episode presents Experian’s defensive case: it positions itself as a behavior-based, regulated data provider trying to empower consumers and lenders with real-time data and governance-minded AI. Lintner stresses human oversight, security-first spending, and consumer tools (Boost, locks) while acknowledging the legitimate fears around scale, bias, and opaque automated decisioning. Whether listeners accept that framing depends on their views about centralized data power and the sufficiency of current governance and legal safeguards.