Navigating Maryland's AI Employment Law: A Guide to AI Impact Assessments

The rapid proliferation of Artificial Intelligence (AI) in business operations, particularly in human resources, presents both transformative opportunities and significant regulatory challenges. As state legislatures increasingly enact specialized AI regulations, organizations must adapt quickly to avoid substantial penalties and reputational damage. Maryland is at the forefront of this movement with its groundbreaking AI Employment Law, Bill HB 1106, which comes into effect on October 1, 2025. For any company using AI in employment decisions, understanding and implementing robust AI compliance software and processes, including comprehensive AI Impact Assessments, is no longer optional—it's imperative.

This article provides compliance officers, general counsel, and CTOs with a detailed, actionable guide on how to conduct an AI Impact Assessment (AIIA) under Maryland's new law. We'll delve into the specific requirements, offer practical steps for implementation, and demonstrate how an AI compliance platform like AICompliant can streamline your efforts, ensuring proactive risk management and seamless adherence to the evolving regulatory landscape.

Maryland's HB 1106: Why It Matters for Your Business

The Maryland AI Employment Law (HB 1106), effective October 1, 2025, marks a pivotal moment for employers leveraging AI in the Free State. This legislation specifically targets the use of AI tools in various employment-related contexts, including hiring, firing, promotion, and other human resources decisions. Its primary objective is to prevent algorithmic discrimination, ensuring fairness and equity in AI-driven employment practices.

Maryland's law requires employers to:

Conduct an Impact Assessment: Before using an AI tool for employment decisions, employers must perform an impact assessment to identify and mitigate potential biases or discriminatory outcomes.
Provide Notice: Employees and job applicants must be notified when AI is used in evaluating their qualifications or performance.
Offer Alternative Accommodations: In certain circumstances, employers may need to provide alternative selection processes if an AI tool disproportionately impacts protected classes.

Non-compliance with HB 1106 can result in significant penalties, with fines reaching up to $10,000 per violation, enforced by the Maryland Commissioner of Labor and Industry. This makes proactive compliance and the deployment of effective AI compliance automation tools critical.

Maryland’s approach mirrors the growing legislative trend seen in other jurisdictions. For instance, the NYC Automated Employment Decision Tool (AEDT) Law (Local Law 144 of 2021), effective July 5, 2023, imposes similar requirements, including bias audits and notice, carrying penalties of up to $1,500 per violation per day. This patchwork of regulations underscores the urgent need for a unified AI risk management framework and a comprehensive US state AI law tracker to manage multi-jurisdictional compliance.

What is an AI Impact Assessment (AIIA)?

An AI Impact Assessment (AIIA) is a systematic process designed to identify, evaluate, and mitigate risks associated with the development, deployment, and use of AI systems. For Maryland's HB 1106, the focus is squarely on AI systems used in employment. The AIIA aims to uncover potential biases, discriminatory impacts, privacy concerns, and other ethical or legal risks that an AI tool might pose, particularly concerning protected characteristics like race, gender, age, or disability.

Conducting a thorough AIIA is essential not only for legal compliance but also for upholding ethical standards, building trust, and safeguarding your organization's reputation. It serves as a cornerstone for robust algorithmic discrimination prevention.

Step-by-Step Guide to Conducting an AI Impact Assessment Under Maryland HB 1106

Performing an effective AIIA requires a structured approach. Here’s a six-step process, incorporating best practices and specific considerations for Maryland’s law:

1. Define Scope and Identify AI Systems in Employment

The first step is to clearly define which AI systems fall under the purview of Maryland HB 1106.

Inventory AI Tools: Create a comprehensive inventory of all AI and machine learning tools currently used or planned for use in employment decisions. This includes tools for resume screening, video interview analysis (e.g., similar to those covered by the Illinois AI Video Interview Act, HB 2557, effective January 1, 2020), performance monitoring, predictive analytics for promotions, or even internal HR chatbots that influence employment outcomes.
Determine AI System Purpose and Scope: For each identified system, document its intended purpose, the specific employment decisions it influences, and the data it processes. For example, is it used for initial candidate screening (high volume, early stage impact) or for final selection recommendations (higher impact, later stage)?
Identify Stakeholders: Assemble a cross-functional team including representatives from HR, legal, IT/AI development, compliance, and potentially external experts.

Practical Tip: Leveraging an AI compliance platform can significantly streamline this inventory process, allowing you to centralize documentation for all AI tools across your organization and track their regulatory relevance.

2. Data Governance and Bias Review

The quality and nature of the data used to train and operate AI systems are paramount to preventing discriminatory outcomes. This step focuses on assessing the data lifecycle.

Training Data Assessment:
- Source and Representativeness: Evaluate the source of training data. Is it representative of the target population for employment decisions? Does it include diverse demographic groups? For instance, California's AB 2013, effective January 1, 2025, specifically addresses training data, with penalties up to $7,500 per violation, underscoring the importance of this aspect.
- Bias Detection: Analyze training data for inherent biases, stereotypes, or historical inequities. This can involve statistical analysis to identify disparities across protected characteristics.
- Data Minimization: Ensure that only necessary and relevant data is collected and processed, aligning with privacy principles like those in the GDPR (Regulation (EU) 2016/679, effective May 25, 2018, with penalties up to $20,000,000 per violation) and US state privacy laws like the Virginia CDPA, effective January 1, 2023.
Data Processing and Usage: Assess how data is processed, transformed, and used by the AI system. Are there steps that could inadvertently amplify existing biases or create new ones?

Practical Tip: Develop an AI impact assessment template that includes specific fields for data source, representativeness analysis, and bias detection methodologies. AICompliant's platform offers built-in templates and data analysis tools to facilitate this critical review.

3. Risk Identification and Mitigation of Algorithmic Discrimination

This is the core of the AIIA for HB 1106, focusing on identifying and addressing potential discriminatory impacts.

Identify Potential Harms: Beyond direct discrimination, consider indirect or disparate impacts. Could the AI system lead to adverse outcomes for certain demographic groups, even if not explicitly programmed to do so? This includes assessing potential impacts on opportunities, working conditions, and privacy.
Fairness Metrics Evaluation: Employ various fairness metrics (e.g., demographic parity, equalized odds, predictive parity) to quantitatively assess if the AI system produces equitable outcomes across different groups. This often requires access to sensitive demographic data (used for analysis, not for decision-making by the AI).
Scenario Testing and Simulations: Conduct "what-if" analyses and simulations to test the AI system's behavior under different conditions and with diverse input data. This helps uncover edge cases or vulnerabilities that might lead to unfair outcomes.
Mitigation Strategies: Develop concrete strategies to mitigate identified risks. This might include:
- Bias Correction Techniques: Implementing algorithmic debiasing methods.
- Feature Engineering: Removing or transforming problematic input features.
- Human Oversight: Incorporating human review points in the AI-driven workflow.
- Retraining or Redesign: If biases are deeply embedded, retraining the model with more balanced data or redesigning the system may be necessary.

Practical Tip: Adopting an AI risk management framework like the NIST AI RMF 1.0 (voluntary, but a powerful standard) provides a structured approach to identifying and managing these risks. An AI compliance tool can help document and track the implementation of mitigation strategies, providing an auditable trail.

4. Transparency and Explainability

Maryland's law, similar to other regulations, emphasizes the need for transparency.

Understand AI Logic: Document the AI system's architecture, algorithms, and decision-making processes to the extent possible. While "black box" models can be challenging, strive for actionable insights into how the AI reaches its conclusions.
Explainability Requirements: Can the outcomes of the AI system be clearly explained to affected individuals (e.g., why a candidate was not selected)? This directly relates to provisions found in frameworks like the OECD AI Principles, effective May 22, 2019, which promote transparent and explainable AI.
Notice to Individuals: Ensure that mechanisms are in place to notify employees and applicants when an AI system is being used in an employment decision, as required by HB 1106. This notice should be clear, concise, and accessible.
Data Subject Rights: Implement processes to handle requests related to data access, correction, and explanation of AI decisions, aligning with broader data privacy rights.

Practical Tip: An AI compliance checklist 2026 should include specific items for transparency disclosures and explainability requirements. AICompliant helps generate and manage these disclosures, ensuring consistency and compliance with regulations like the California AI Transparency Act (SB 942), effective January 1, 2026, which carries penalties up to $5,000 per violation per day.

5. Documentation and Reporting

Robust documentation is crucial for demonstrating compliance, especially in the event of an audit or complaint.

Record All Assessments: Maintain detailed records of every AIIA conducted, including the scope, methodologies used, findings (identified risks and biases), mitigation strategies implemented, and the rationale behind decisions.
Policy Development: Establish clear internal policies and procedures for the responsible use of AI in employment, incorporating the findings of your AIIAs.
Stakeholder Communication: Document communications with relevant stakeholders, including legal, HR, and external experts.
Reporting: Prepare internal reports for senior management and the board, outlining AI risks, compliance status, and mitigation progress. This demonstrates effective governance.

Practical Tip: An AI compliance platform like AICompliant provides a centralized dashboard for all compliance activities, allowing you to store documentation, track policies, and generate audit-ready reports effortlessly. This is invaluable for demonstrating adherence to Maryland HB 1106 and other regulations.

6. Continuous Monitoring and Review

AI systems are not static; they learn, evolve, and can degrade over time (model drift). Compliance is an ongoing process.

Regular Re-assessment: Establish a schedule for periodic re-assessment of AI systems, especially after significant updates, changes in usage, or when new data sources are introduced. Maryland's law implies an ongoing obligation to ensure fairness.
Performance Monitoring: Continuously monitor the AI system's performance, paying close attention to fairness metrics and potential disparities in outcomes over time.
Incident Response: Develop a plan for responding to identified biases or discriminatory outcomes. This includes root cause analysis and implementing corrective actions promptly. California's SB 53 (Frontier AI / Incident Reporting), effective September 29, 2025, with penalties up to $1,000,000 per violation, highlights the increasing focus on AI incident reporting.
Stay Updated: Regularly review changes in AI regulation and best practices (e.g., ISO/IEC 42001:2023, effective December 18, 2023) to ensure your AIIA process remains current and effective.

Practical Tip: An AI compliance tool with continuous monitoring capabilities is essential here. AICompliant's automated alerts and real-time dashboards can flag potential issues, allowing your team to proactively address them. You can use the AICompliant /tools/compliance-checker regularly to assess your ongoing adherence.

Leveraging AICompliant for Your Maryland HB 1106 Compliance

For companies grappling with the complexities of Maryland's AI Employment Law and the broader landscape of AI regulations, an advanced AI compliance software solution like AICompliant is indispensable. Our platform is specifically designed to transform the daunting task of AI compliance into a streamlined, manageable process.

AICompliant empowers your organization to:

Automate AI Impact Assessments: Our platform offers guided workflows and templates tailored to specific regulations, including Maryland HB 1106. This significantly reduces manual effort and ensures all necessary steps are covered.
Centralize AI System Inventory: Maintain a single, auditable record of all AI systems, their purposes, data inputs, and associated risks, acting as your definitive US state AI law tracker.
Facilitate Algorithmic Bias Detection: Integrate tools for data analysis and fairness metric calculation, aiding in the proactive identification and mitigation of algorithmic discrimination.
Streamline Documentation and Reporting: Automatically generate comprehensive reports and maintain an immutable audit trail, essential for demonstrating compliance to regulators like the Maryland Commissioner of Labor and Industry. Our /dashboard provides real-time insights into your compliance posture.
Enable Continuous Monitoring: Set up alerts and automated checks to monitor AI system performance and compliance status, ensuring your AI systems remain compliant even as they evolve.
Adapt to Evolving Regulations: With pre-built modules for various global and national AI laws, AICompliant helps you stay ahead of the curve, managing not just Maryland's law but also the Colorado AI Act (SB 24-205, effective June 30, 2026, with penalties up to $20,000 per violation), the EU AI Act (Regulation (EU) 2024/1689, effective August 1, 2024, with high-risk system obligations enforceable from August 2, 2026, and penalties up to $35,000,000 per violation), and more.

By leveraging AICompliant's AI compliance automation capabilities, your team can move beyond manual spreadsheets and fragmented processes. You gain the confidence that your AI deployments are fair, transparent, and fully compliant, safeguarding your business from regulatory penalties and fostering responsible innovation.

The Broader Landscape of AI Regulation

While Maryland's HB 1106 focuses on employment, it's crucial to recognize it as part of a rapidly expanding global regulatory framework. The demand for robust AI compliance solutions extends beyond state lines.

European Union: The EU AI Act, a landmark regulation, categorizes AI systems by risk, imposing stringent requirements, especially on "high-risk" applications like those in employment, with enforcement for high-risk systems beginning August 2, 2026.
Other US States: Beyond Maryland and New York City, states like Utah (SB 149, effective May 1, 2024, penalties up to $10,000), Connecticut (SB 1103, effective October 1, 2025, penalties up to $5,000), and Texas (Texas Responsible AI Governance Act, HB 149, effective January 1, 2026, penalties up to $200,000) are enacting their own distinct AI policies.
Federal Enforcement: The FTC, under Section 5 of the FTC Act, has increasingly focused on AI-related deception and unfair practices, signaling potential enforcement actions with penalties up to $50,000 per violation per day.
International Standards: Frameworks like NIST AI RMF and ISO/IEC 42001 provide essential voluntary guidelines that are being incorporated into regulatory requirements worldwide.

This complex and evolving landscape necessitates a strategic, long-term approach to AI governance. An integrated AI compliance platform is not just a tool for today's compliance but an essential partner for navigating the future of AI regulation.

Conclusion: Proactive Compliance is Your Best Defense

The Maryland AI Employment Law (HB 1106), effective October 1, 2025, signals a new era for AI governance in the workplace. For mid-to-large companies operating in Maryland or considering AI in their HR functions, conducting thorough AI Impact Assessments is a non-negotiable step toward responsible AI adoption and legal compliance. By systematically identifying and mitigating risks of algorithmic discrimination, you not only adhere to the law but also strengthen your commitment to fairness and ethical AI.

Embrace proactive compliance. The financial penalties and reputational damage associated with non-compliance far outweigh the investment in robust AI governance. With an advanced AI compliance software solution like AICompliant, you can confidently navigate these complexities, automate your compliance workflows, and ensure your AI initiatives contribute positively to your organizational values and objectives.

Unlock Seamless AI Compliance.

Don't let complex AI regulations like Maryland's HB 1106 hinder your innovation. Discover how AICompliant's intelligent platform can automate your AI Impact Assessments, manage your compliance obligations, and safeguard your business.

Explore AICompliant Pricing & Solutions Today

AI Compliance Software: Maryland AI Impact Assessment Guide