Intermedia uses AI for redaction of call center transcripts

Consumer data privacy continues to be a challenge for contact centers, whether they are hosted internally by companies or outsourced to places like India, especially when so much personal data ends up in call center transcriptions that are intended to help these centers to better support callers.

But Intermedia Cloud Communications may have an answer to keeping the private data that passes through call centers private, with an AI-driven capability for redacting sensitive data from call center transcriptions called AI Transcription Redaction. The new offering is powered by Intermedia’s SPARK AI technology, and is designed to automatically identify and redact sensitive customer information from Intermedia Contact Center voice transcriptions.

Intermedia has been launching a growing number of AI-based call center features, including AI Interaction Insights, a suite of features which allows Intermedia’s omnichannel Contact Center customer to take advantage of AI-enabled transcription of customer calls, allow customer support personnel access to call summaries, the ability to flag interactions for review based on key words and phrases, and the opportunity for instant sentiment analysis of each call. The new AI Transcription Redaction feature is available at no cost to Intermedia Contact Center customers as the latest addition to the AI Interaction Insights suite.

Irina Shamkova, executive vice president of product management at Intermedia, told Fierce Electronics via email that Intermedia has always adhered to strict security and privacy protocols in handling and storing data, but that call transcriptions were not subject to redaction. “The implementation of our AI Transcription Redaction feature changes this, ensuring sensitive information in all call recordings is redacted by default,” she said. “Other Intermedia services, including Archiving, AI Summarization, and other AI Insights, will utilize these redacted transcripts. This can only be overridden by an account administrator, thus providing an additional layer of control.”

The new offering comes at a time when generative AI is enabling a wide range of new capabilities for many companies, but concerns remain that AI-enabled features could be misused or allowed to commit unethical behavior. In this case, Shamkova said other AI-powered applications would not be able to access data that has already been redacted.

“In terms of concerns about the accessibility of redacted data by other generative AI systems, once data is redacted, it is effectively and permanently removed from transcripts, making it inaccessible to any subsequent processes or systems, AI-based or otherwise,” she said. “Our primary goal with this technology is to ensure the utmost privacy and security of our customers' data.