For years, the insurance industry treated artificial intelligence risk like an awkward dinner guest—present but unacknowledged, tucked into existing policies without anyone saying the words out loud. That studied avoidance is crumbling.
Starting in early 2026, a handful of major carriers and Lloyd's syndicates began rolling out what the industry calls "affirmative AI liability" products. These are explicit policies designed to cover what happens when algorithms misfire, discriminate, hallucinate, or simply fail in ways that cost companies money and reputation. The shift represents something of a forced reckoning. ISO, the insurance industry's influential standards body, introduced AI-related exclusions that took effect in January 2026, and established carriers—W. R. Berkley, Great American, AIG among them—filed their own carve-outs in lines like directors and officers liability and professional liability. The message to corporate risk managers was blunt: if you want AI risk covered, you'll need to ask for it by name.
What followed was a quiet but telling scramble.
Munich Re Breaks the Ice
On March 18, HSB—a Munich Re company perhaps better known for insuring boilers and industrial machinery—announced its AI Liability Insurance product. The timing mattered. With ISO exclusions in place and "silent AI" exposures vanishing from standard policies, HSB saw an opening that few institutional players had yet claimed.
The product targets business interruption, bodily injury, property damage, and advertising injury stemming from AI incidents. That's a wide net. But what really distinguished the launch was the institutional backing. Munich Re's balance sheet provides significant capacity that can scale well beyond the experimental Lloyd's programs or niche managing general underwriters that have historically populated the frontier of strange new risks.
Cowbell and Lloyd's Move In
Less than five weeks later, on April 21, Cowbell—a cyber insurance upstart that's been pitching itself as the modern answer to legacy carriers—launched Prime One in the United States. The product explicitly covers emerging AI and quantum risks (quantum, apparently, is the new aspirational risk category). Limits run up to $10 million, and the target market is organizations with $250 million to $1 billion in revenue. These are enterprises large enough to deploy sophisticated AI systems, wealthy enough to attract litigation, and savvy enough to know they might need something beyond standard cyber policies.
Lloyd's of London, predictably, had already been tinkering. On February 10, Chaucer—a Lloyd's syndicate—partnered with risk platform Armilla to launch Vanguard AI, which coordinates cyber, technology errors and omissions, and AI-specific coverage through a single underwriting framework. Lloyd's, after all, has always been the market's laboratory for exotic and emerging risks. The structure here is interesting: rather than offering standalone AI policies, Chaucer stitched together multiple coverage lines, betting that AI incidents rarely fall neatly into a single category.
Counterpart, a smaller player in the specialty liability space, had actually tested the waters even earlier—on November 24, 2025. That product added a technology E&O insuring agreement addressing hallucinated reports, misclassified exposures, and flawed hiring recommendations. The bread-and-butter claims, in other words, likely to emerge as enterprises lean harder on generative AI and discover that it occasionally fabricates things with alarming confidence.
What Actually Gets Covered (and What Doesn't)
The products launched so far share common threads. They cover AI-generated errors, bias and discrimination claims, outputs that cause harm, copyright and intellectual property disputes, and privacy injuries. Most offerings attach as endorsements to existing cyber, tech E&O, or professional liability policies rather than standing alone—though the Lloyd's-backed programs are exceptions, reflecting Lloyd's preference for bespoke solutions.
One industry roundup from early May noted that Corgi, a managing general underwriter, now offers AI insurance coverage as an endorsement rather than a standalone policy. That's becoming the pattern: insurers are stitching AI protection into familiar structures rather than inventing entirely new product categories. It's faster, simpler, and probably smarter from a regulatory standpoint.
The coverage isn't unlimited, of course. Underwriters are scrutinizing governance artifacts closely. Companies adopting NIST AI Risk Management Framework controls or ISO 42001-style standards will find it easier to bind affirmative coverage. Carriers want to see that policyholders are managing AI deployments—treating them as engineered systems with failure modes—not as magic boxes that spit out answers.
That's a significant shift. For insurers, the existence of governance frameworks transforms AI from an unquantifiable existential risk into something that looks more like industrial process risk, which they've been underwriting for decades.
Why the Sudden Urgency
Gartner issued guidance in early April advising general counsel to assess AI insurance specifically, forecasting growing AI-related claims. Around the same time, Aon published its AI Fact Sheet mapping AI perils across insurance lines and highlighting the prevalence of "silent AI" exposures—those unintentional gaps where coverage may or may not exist, depending on how a judge reads a policy written before anyone had heard of large language models.
An April trade publication article declared that "AI liability emerges as the new cyber for SMEs," capturing how quickly the risk has moved from theoretical to urgent. Small and midsize enterprises deploying AI—often without the compliance infrastructure of Fortune 500 companies—represent a vast, underinsured market. They're also the ones most likely to stumble into trouble, lacking the legal and technical teams to vet AI tools before deployment.
Earlier reporting from S&P Global Market Intelligence noted that as insurers retreat from unintentional AI risk through exclusions, startups and specialty syndicates see opportunity. That's the pattern whenever new risks emerge: established carriers pull back, and nimble entrants rush in.
Institutional Capital Changes the Game
The presence of Munich Re and Lloyd's syndicates signals something more durable than a speculative trend. Reinsurers—the insurers of insurers—are willing to back AI liability coverage with meaningful capacity. That's a turning point. Early AI liability experiments often relied on tiny Lloyd's programs or captive arrangements with limited capital behind them. HSB's March launch, backed by one of the world's largest reinsurers, suggests the market believes AI risk is quantifiable and insurable at scale.
Whether that belief is justified remains to be seen. The actuarial data on AI failures is still thin, and the legal landscape is evolving. But the willingness to deploy capital suggests insurers think they can price the risk, which is half the battle.
For now, affirmative AI coverage remains concentrated among enterprises large enough to navigate specialty markets and willing to pay for explicit protection. But as the products mature and underwriting models sharpen, the coverage will likely trickle down to smaller organizations. Especially as regulators in the U.S. and Europe begin mandating AI governance standards—which they inevitably will—demand for affirmative coverage should accelerate.
The era of pretending AI risk doesn't exist, or that it's covered somewhere in a general liability policy, is over. The question now is how fast enterprises will recognize the gap—and how much they'll pay to fill it. If history is any guide, most will wait until after the first big lawsuit.
