Risk Management May 26, 2026 7 min read

Cyber Liability Insurance and AI: What Underwriters Are Now Asking For

Cyber liability insurance renewals in 2026 look different from two years ago. Underwriters are asking specific questions about AI governance, and the answers — or the absence of them — are affecting coverage terms, premiums, and in some cases, coverage availability. Here's what to expect and how to be prepared.

Note: Insurance underwriting criteria vary significantly by carrier, policy type, and industry. Consult your insurance broker regarding your specific coverage situation and how AI governance factors into your particular policy.

Cyber insurance has been evolving rapidly in response to rising claim volumes. After years of expanding coverage and relatively flat premiums, the market tightened significantly as ransomware claims increased. Now, AI is adding a new dimension to the underwriting conversation — one that most companies aren't prepared for.

The underlying concern from underwriters is straightforward: shadow AI creates data exposure risk that is invisible to the insured company, which makes it invisible in the underwriting model, which makes it a source of unexpected claims. Documented AI governance is how insureds demonstrate they've thought about this risk and taken steps to address it.

$670k average additional cost per breach when shadow AI is involved — the primary driver being detection time, since security teams have no visibility into unsanctioned AI sessions. IBM Cost of a Data Breach Report, 2025

The AI governance questions appearing on renewal applications

These questions are appearing on cyber insurance applications and renewal questionnaires from multiple major carriers. Not every carrier asks every question — but the trend is consistent across the market, and the questions are becoming more specific each renewal cycle.

Underwriter question
Does the organization have a written policy governing employee use of AI tools?
What a strong answer looks like:

Yes — we have a written AI Acceptable Use Policy covering all employees and contractors. The policy was last reviewed in [date] and covers approved tool classifications, data handling restrictions, human oversight requirements, and incident reporting. All employees have formally acknowledged receipt of the policy.

Underwriter question
Does the organization restrict which AI tools employees may use for work purposes?
What a strong answer looks like:

Yes — we maintain a tool tier list that classifies AI tools into approved, limited-use, and prohibited categories. Employees are required to check the tier list before using a new AI tool and to seek approval through a defined process for tools not yet classified. The tier list is updated quarterly or when significant new tools emerge.

Underwriter question
Does the organization restrict what types of data employees may share with AI tools?
What a strong answer looks like:

Yes — our policy specifies data categories that may not be shared with any AI tool without specific approval, including customer PII, non-public financial information, confidential business strategy, and employee personal data. Approved tool tiers also specify which data categories can be shared with each tier.

Underwriter question
Have employees been trained on AI governance expectations?
What a strong answer looks like:

Yes — all employees received a policy walkthrough when the AI policy was launched. Managers received additional briefing on enforcement and escalation. Acknowledgment of the policy was required of all employees within 30 days of launch, and is part of new employee onboarding.

Underwriter question
Has the organization evaluated the data handling practices of AI vendors used by employees?
What a strong answer looks like:

Yes — for AI tools on our approved list, we have reviewed data processing terms and confirmed whether data processing agreements or no-training commitments are in place. This assessment was the basis for our tier classifications. We require data processing agreements for any AI tool approved for use with sensitive data categories.

The gap between good and bad answers

The difference in how underwriters respond to these questions is stark. Here's what that looks like for the core governance question:

Weak answer

"We've told employees not to use unapproved AI tools for sensitive work, but we don't have a formal written policy yet."

Strong answer

"Yes, we have a written AI Acceptable Use Policy with a tool tier list. All employees signed acknowledgment forms in Q1 2026."

The weak answer signals an unquantified exposure. The strong answer signals a governed, documented risk position. Underwriters price the difference.

What "better terms" actually means

AI governance documentation doesn't guarantee lower premiums — the insurance market is more complex than that, and premiums are affected by many factors beyond AI. What documented AI governance does affect:

Coverage availability in regulated industries. Some carriers are excluding AI-related claims or adding AI-specific sublimits for insureds that can't demonstrate basic AI governance. If your renewal questionnaire includes AI questions and you can't answer them, you may find coverage narrowing.

The underwriting conversation itself. Brokers and underwriters report that companies with documented AI governance programs have materially better renewal conversations — not necessarily dramatically lower premiums, but fewer coverage restrictions, faster underwriting, and a relationship of demonstrated risk management rather than demonstrated exposure.

Claims handling. If an AI-related breach occurs and it emerges that the company had no AI governance program, claims handlers and coverage counsel will scrutinize whether relevant coverage exclusions apply. A documented governance program is evidence that the company acted responsibly — relevant if a coverage dispute arises.

"Cyber liability carriers are paying attention to AI governance: many now require documented AI policies for favorable coverage terms or are adding AI-specific exclusions for companies without them." — Jonathan Lasley AI Advisory, 2026

The three documents that answer the underwriting questions

The good news: answering the AI governance underwriting questions well doesn't require an enterprise compliance program. It requires three documents that most mid-market companies can have in place within a few weeks:

These three documents together constitute a defensible AI governance program for the purposes of cyber insurance underwriting. They don't require a legal team, an IT department, or a six-month project.

What to do before your next renewal

Get renewal-ready AI governance documents in 10 minutes.

Shadow AI Policy generates the three documents that answer cyber insurance underwriting questions — acceptable use policy, tool tier list, and employee acknowledgment form — tailored to your industry and company size.

Generate my policy kit →