also with regard to the timing the policy will be due. Unlike the fear of terrorism and other grand-impact events, such as global health pandemics that will affect many policyholders with identical damages at once, the AI market is far more diversified and specified.Thepre-existingcategories ofspecialized insurancepoliciesallow insurers to issue policies to AI users, manufacturers, or whoever isobligated or desires to purchase this type of hedging within their field of expertise. An important caveat to the distribution element is the possibility of catastrophic risk. Although AIs will conduct many of the same activities carried out by individuals, there is a greaterrisk of aggregation harm because AIs run on common technologies and may be subject to common risks. This very issue has hampered the development of the cyber insurance market, given that widespread damages may, and do, occur when many companies use thesame online platforms to store their data. 216Though this issue cannot be completely eliminated, it seems less significant in the AI context because AI can operate on different platforms in light of their different functions and aims, which are not limited to datastorage as in the cyber context. Furthermore, this issue can be mitigated by manufacturers ensuring their products operate on different and protected platforms, thereby distributing risks, to prevent wide-spread damages. Such mitigation can be encouraged byinsurers who have the incentive to prevent these types of aggregate disasters.Third, isthe element ofprofitability. While meeting the two first elements is sufficient to make a risk insurable, lacking the third element essentiallymeans that there will be no market for insuring these types of risks. 217If profitability is not achievable, but there isstill ademand for insurance,the government willusuallystep in with government-subsidized premiumsto bridge this gap. This government support allows insurers to offer policiesto the public and still make a profit. In the AI context, theprofitability of policies covering commercial AI damagesshould not be an