Maryland AI Liability Insurance

The Shift from Traditional Professional Liability to AI-Specific Coverage

Traditional technology errors and omissions policies were built around human professional judgment: a consultant gives bad advice, a developer ships buggy code, an IT provider causes downtime. AI liability is structurally different. The "error" may originate from a model trained on data your company never reviewed, producing an output no human approved before it reached the end user.

Standard tech E&O forms often contain exclusions for automated decision-making or for outputs not directly supervised by a named insured. A policy form designed for AI risk will include an insuring agreement that specifically addresses model outputs, training-data defects, and the gap between intended and actual system behavior. The distinction is not academic: it determines whether your carrier owes you a defense when a plaintiff's complaint arrives.

Why Baltimore and Columbia Tech Hubs Face Unique Risk Profiles

Baltimore's growing cybersecurity and health-tech corridors place companies at the intersection of AI deployment and heavily regulated industries. If your AI tool touches protected health information or financial data, you face overlapping federal and state enforcement exposure. Columbia's concentration of defense contractors and government-adjacent technology firms adds another layer: federal procurement rules increasingly require AI risk disclosures, and a bias claim against a government subcontractor can trigger both civil liability and contract termination.

Annapolis-based firms, particularly those serving state government, must account for Maryland-specific procurement standards and the state's emerging AI transparency requirements. Your insurance program should reflect these jurisdictional realities, not generic national assumptions.

Protecting Against Financial Loss from Inaccurate Data Generation

A policy form that responds to AI hallucination risk will typically cover third-party claims arising from financial loss caused by inaccurate AI-generated content. The key language to examine is whether the insuring agreement covers "wrongful acts" broadly enough to include automated outputs, or whether it limits coverage to acts performed by a natural person.

You should also check for sublimits on data-accuracy claims. Some forms cap these at $250,000 or $500,000, well below the potential exposure if your AI system feeds incorrect data into a client's financial model or regulatory filing. The retention, or deductible, matters too: a $50,000 retention on a $250,000 sublimit leaves you with very little net recovery. This is exactly the kind of form-level detail that a specialist agency like Bloc Cyber reviews before binding, because the gap between the declarations page and the actual coverage grant is where claims go unpaid.

Legal Defense for Defamation and Intellectual Property Claims

AI-generated content that reproduces copyrighted material or publishes false statements about an identifiable person creates defamation and IP infringement exposure. Generative AI liability policies are beginning to include specific insuring agreements for intellectual property infringement arising from model outputs. You need to confirm whether your form covers both defense costs and indemnity for IP claims, and whether the definition of "advertising injury" in your policy extends to AI-generated content or is limited to traditional media.

Compliance with Maryland Fair Employment and Housing Laws

Maryland's anti-discrimination statutes apply to AI-driven decisions just as they apply to human decisions. If your hiring algorithm screens out candidates in a pattern that correlates with race, gender, age, or disability, you face exposure under both state and federal civil rights laws. General counsel offices are now being advised to assess AI insurance as a core mitigation strategy for exactly this category of risk.

Your policy form should not exclude employment-related claims entirely, which many standard professional liability forms do. A purpose-built AI liability form will carve back coverage for algorithmic bias in employment decisions, subject to appropriate retentions and limits.

Coverage for Regulatory Fines and Settlement Costs

Regulatory fines present a particular challenge because many insurance forms exclude "fines and penalties" by default. A well-structured AI liability policy will include an insurable-fines carve-back, covering civil penalties and regulatory assessments where permitted by law. Maryland law generally permits the insurance of civil fines, but the policy form must affirmatively include them. Settlement costs for class-action bias claims can easily reach seven figures for a mid-market company, so your per-claim and aggregate limits need to reflect realistic exposure, not minimum premium thresholds.

Liability for Unintended Actions by AI Agents

When an AI agent autonomously cancels a customer's account, submits a regulatory filing, or executes a trade, the company deploying that agent bears the liability for the outcome. Autonomous AI systems can void cyber insurance coverage if the policy form was not written to contemplate machine-initiated actions. You need to verify that your policy's definition of "insured operations" or "professional services" includes actions taken by autonomous systems on your behalf.

The high-tech insurance risk landscape in 2026 reflects this shift: carriers are writing endorsements that specifically address agentic AI, but these endorsements vary widely in scope. Some cover only "supervised autonomy," requiring evidence that a human could have intervened. Others extend to fully autonomous operations.

Setting Appropriate Limits for High-Autonomy Systems

Limit adequacy for agentic AI depends on the volume and value of decisions the system makes. A customer service chatbot that can issue refunds up to $100 presents a different exposure profile than an AI agent that approves six-figure purchase orders. Your limits should reflect the maximum plausible single-event loss and the aggregate exposure over a policy period.

A common mistake is purchasing a $1 million aggregate limit for an AI system that processes thousands of autonomous decisions daily. If a systemic error affects a large number of transactions before detection, the aggregate can exhaust in a single incident. Bloc Cyber's approach to AI and algorithmic liability placement focuses on matching limits and retentions to the actual decision volume and financial authority of each deployed system.

Does my existing cyber liability policy cover AI-related claims? It may not. Many cyber forms now contain explicit AI exclusions, and even those without exclusions may not affirmatively cover AI outputs. You need to read the insuring agreements and exclusions, not the marketing summary.

Is AI liability insurance required by Maryland law? No state mandate exists as of 2026, but regulatory enforcement actions and private lawsuits create strong financial incentives to carry coverage. Companies in healthcare and financial services face the highest exposure.

What triggers a hallucination-related claim? A third party suffers a financial loss or reputational harm because they relied on an AI-generated output from your system that contained fabricated or materially inaccurate information.

Can I add AI coverage to my existing tech E&O policy? Some carriers offer endorsements, but AI risk is evolving faster than many existing policy forms can accommodate. A standalone AI liability form often provides broader and clearer coverage.

How are premiums determined for AI liability insurance? Carriers evaluate the type of AI deployed, the degree of autonomy, the industry, the volume of AI-driven decisions, and your internal governance and testing protocols. Errors and omissions premiums for technology firms provide a baseline, but AI-specific factors can adjust pricing significantly.

The gap between what Maryland businesses assume their insurance covers and what the policy form actually says about AI risk is real and measurable. Whether your exposure centers on hallucination errors, algorithmic bias, or agentic AI decisions, the coverage response depends entirely on how the insuring agreements, exclusions, and sublimits are written.

Your next step is straightforward: pull your current policy form, check for AI-related exclusions, and compare the insuring agreements against the specific AI systems you have deployed or plan to deploy. If the language does not clearly address your exposure, or if you are not sure how to read the form, request a coverage review so a specialist can walk through the policy with you and identify where the gaps sit before a claim finds them first.

Cyber Coverage Built for Your Industry

Every industry faces a different cyber threat, from patient records in healthcare to wire fraud in construction. Bloc Cyber matches coverage to the risks your sector actually faces, drawing on specialty carriers that understand your business. Find your industry below to see how we protect it.

Field notes from the placement desk.
What carriers are asking right now.