In the example illustrated in FIG. 1, the communication service 108 obtains a trust score from a scoring agent 110 hosted on a server 106b in the service provider environment 104. The scoring agent 110 may be a service or module configured to generate a trust score for a customer account using cost information associated with connecting to a destination endpoint and information linked to the customer account that impacts a decision whether to allow a voice connection to the destination endpoint. For example, the scoring agent 110 may receive a request for a trust score from the communication service 108 that specifies a customer account and a destination endpoint (e.g., a phone number or other destination endpoint identifier which can be assigned to a destination device 134 or 136). In response to the request, the scoring agent 110 may use the destination endpoint to obtain destination data 112, and the customer account to obtain customer data 116, and call history data 114. Destination data 112 may include a destination rate (e.g., standard rate, premium rate, or international premium rate), a destination endpoint type (e.g., landline, cellular phone, or satellite phone), a destination trust indicator of known or suspected fraudulent activity, and other destination data 112. Customer data 116 may include a customer account status, customer account type, billing account balance, payment history, customer account size, age of customer account, history of fraudulent activity, and other information linked to a customer account. Call history data 114 may include historical data associated with calls linked to a customer account, including patterns of connections to destination endpoints, volume of calls made to destination endpoints, and other data related to calls made to destination endpoints. After obtaining the data 112/114/116, the scoring agent 110 may use the data 112/114/116 to generate a trust score for the customer account, as explained below, and return the trust score to the communication service 108.