There's a new line in many commercial general liability policies, and most Illinois business owners don't know it's there.
ISO, the organization that develops the standardized forms most carriers use, filed two AI-specific exclusion endorsements with state regulators in January 2026: CG 40 47 and CG 40 48. Carriers are adopting them. AIG filed AI exclusion language with the Illinois Department of Insurance specifically. Great American and W.R. Berkley have filed their own AI exclusions as well.
If your GL policy renewed in 2026 and you haven't looked at the endorsement schedule, you might have one of these exclusions without knowing it.
What ISO CG 40 47 and CG 40 48 actually say
These two endorsements aren't identical, and the difference matters.
CG 40 47 is the broader exclusion. It removes coverage for bodily injury, property damage, personal injury, and advertising injury that arises from "AI-generated content" or "AI-enabled systems." The definitions are broad enough to reach most tools businesses are deploying right now, including large language models, automated decision tools, and machine learning systems embedded in business software.
CG 40 48 is narrower. It targets only the personal injury and advertising injury coverage within the GL policy. That's the coverage that responds when a business is sued for defamation, false advertising, or copyright infringement. CG 40 48 removes that coverage specifically when the alleged harm traces back to AI-generated content. It doesn't touch bodily injury or property damage.
So if your carrier adopted CG 40 47, you're looking at a materially wider gap. If they took CG 40 48, the exclusion is more targeted but still carves out real exposure most businesses haven't thought about.
Why this happened in January 2026
ISO doesn't file new exclusion forms without a reason. The January 2026 filings are an acknowledgment that standard GL forms weren't designed for AI and that the existing coverage terms were creating real uncertainty.
Carriers had been dealing with ambiguity. A business deploys a chatbot. The chatbot says something defamatory about a competitor. The competitor sues. Does the GL's advertising injury provision respond? The answer under the old forms was genuinely unclear, and unclear coverage terms end up in litigation between the carrier and the policyholder at the worst possible time.
The ISO endorsements resolve the ambiguity in favor of the carrier. AI-related harm is explicitly outside the standard GL unless the policy specifically provides for it.
From an underwriting standpoint, that's rational. GL pricing is built on decades of loss data for physical accidents and traditional advertising disputes. AI is a different risk profile with no comparable actuarial data. The exclusion buys time to understand the exposure before carriers try to price it.
But from a business standpoint, the result is a coverage gap that's now formalized and in writing.
What kinds of businesses this actually hits
The endorsements aren't a problem for every business. If your operation doesn't use AI in any customer-facing or client-service context, the practical impact may be limited. But several situations create real exposure.
Customer service chatbots. If your business uses an AI assistant to answer questions, handle complaints, or give product recommendations, and that chatbot causes harm, CG 40 47 is designed to exclude it. The harm doesn't have to be dramatic. Incorrect product information that causes injury, or a chatbot response that amounts to defamation, could both trigger a GL claim that the exclusion now pulls outside coverage.
AI-generated marketing. This is exactly what CG 40 48 targets. If your team uses AI to generate ad copy, social media posts, or promotional materials, and that content infringes a copyright or makes a false claim about a competitor, you no longer have the advertising injury coverage you probably assumed you had.
Consultants and professional service firms. If you're an accountant, architect, engineer, or management consultant, and AI tools contribute to your work product, professional liability (E&O) is your primary coverage for client claims. But if third parties are harmed, or if there's a bodily injury dimension, GL can become relevant. The exclusions complicate that picture.
Manufacturers and contractors using AI in operations. Companies using AI for quality control, process monitoring, or equipment decisions should look at CG 40 47 carefully. An AI-enabled system that contributes to a workplace injury or property damage incident could trigger exactly the coverage the endorsement removes.
And the exposure isn't limited to Naperville or the Chicago suburbs. Any Illinois business, from a DuPage County professional services firm to a downstate manufacturer, that uses AI in operations is dealing with this.
AIG filed with Illinois specifically
When a carrier files an endorsement with the Illinois Department of Insurance, it applies to policies written or renewed in Illinois after the filing is approved. AIG's filing with Illinois means policyholders renewing AIG commercial GL coverage in this state need to check whether the AI exclusion was added at renewal.
AIG isn't alone. Great American and W.R. Berkley have filed their own AI exclusions. The number of carriers adopting some version of AI-specific exclusion language is growing, and state-specific filings like the one with the Illinois DOI make those changes legally effective for policies issued here.
The implication: if you renewed your GL in 2026, there's a real possibility an AI exclusion was added without anyone calling to tell you about it. Your premium probably didn't change. Your endorsement schedule did.
How to find out if your policy has one
Ask your agent for the endorsement schedule from your current GL policy. It's a list, usually a few pages, of every endorsement that modifies your base policy. You're looking for any endorsement with "artificial intelligence" in the title or an ISO form number in the CG 40 4x range.
If you find one, read it. Specifically:
- Does it exclude all AI-related harm broadly, or only AI-generated content in advertising and personal injury contexts?
- Does it apply to AI your business uses, or also AI used by third parties?
- Is there a carve-out for AI output that a human reviews and approves before it goes to a customer?
That last question matters more than most policyholders realize. Some carriers filed AI exclusions that explicitly preserve coverage when a human reviews and takes responsibility for AI-generated content before use. Others didn't include that carve-out. If your business has a review process, the presence or absence of that carve-out determines whether your process actually preserves your coverage. It's not something you'd know by looking at the premium page.
What the exclusions don't touch
These endorsements don't gut your GL policy entirely. They target specific coverage buckets.
CG 40 48 leaves bodily injury and property damage coverage completely intact. If someone gets hurt on your premises, or your contractor damages a client's building, your GL still responds. The exclusion only cuts into advertising and personal injury coverage.
Even CG 40 47, the broader exclusion, doesn't strip your entire GL. Physical accidents with no AI component are still covered. And most policy forms still provide defense costs even when the underlying claim involves AI, unless the exclusion is written so broadly that it pulls defense costs along with indemnity. That's worth confirming for your specific form, because it varies.
The exclusions carve out specific coverage for specific categories of harm. Outside those categories, most of your GL coverage stays where it was.
Where this fits in the wider AI insurance picture
The GL exclusions don't exist in isolation. They're part of a pattern.
E&O policies are the natural home for AI service errors and professional mistakes. But some E&O forms are adding AI exclusions for the same reasons. Cyber policies have funds transfer fraud provisions that weren't built for AI-directed payment fraud. And now GL policies are formalizing AI exclusions in the standard forms.
If your business is deploying AI in customer-facing, professional, or operational contexts, there's a real possibility your commercial program has more than one gap worth mapping. The GL exclusion is just the one that's most likely to show up on a policy you've had for years and assumed was unchanged.
A handful of specialty carriers are writing AI liability coverage designed specifically for businesses that use AI tools, as opposed to companies that build AI products. That coverage isn't in the standard market yet, but it exists and it's worth knowing about if your business has meaningful AI deployment in customer-facing or financial operations.
The practical step that takes about 20 minutes
Pull your current GL policy's endorsement schedule. Search for "artificial intelligence" or look for form numbers in the CG 40 4x series. Call your agent if you find something and ask them to walk through what the endorsement actually does to your coverage given how your business uses AI.
Most commercial buyers skip this step because it's easy to assume that policies renewed without comment are policies renewed without change. That assumption was reasonable before January 2026. It isn't anymore for businesses using AI in their operations.
If you're an Illinois business and want someone to actually read your GL endorsement schedule and tell you what you have, reach out to Jack Ray directly. He handles commercial accounts across the state and can identify which AI exposures your current program covers and which ones it doesn't.
Email: jray@lakeshoreriskadvisors.com