New capabilities cover 100% of calls for quality and compliance, and protect every live interaction against synthetic caller fraud, real-time social engineering, and agent identity fraud

BERKELEY, Calif.: Krisp, the Voice AI company processing more than 80 billion voice minutes monthly, today announced two new capabilities for contact centers: Voice Security and Speech Analytics. Together, they give contact centers full-spectrum governance of the voice channel: the security to protect every interaction in real time, and the intelligence to know what is happening on every call.

The Threat Landscape

Most contact center quality programs review less than 2% of calls manually. The voice channel in contact centers is the financial control point-the place where customer identity is verified, high-value transactions are authorized, and sensitive data is accessed.

At the same time, voice fraud is growing at a rate existing tools were not designed to address:

• Deepfakes grew from 0.1% to 6.5% of all detected fraud attempts in three years, a 2,137% increase (Signicat, 2024)
• Human agents only identify synthetic voices approximately 60% of the time, approaching chance, with no meaningful improvement from training (UC Berkeley / Nature Scientific Reports, 2025)
• AI and voice cloning contributed to more than $30 million in documented BEC losses in the U.S. in 2025, reported cases only (FBI IC3, 2025)
• Gen-AI-enabled fraud losses projected to reach $40 billion in the U.S. by 2027, up from $12.3 billion in 2023 (Deloitte Center for Financial Services, 2024)

These two gaps, the coverage gap in quality assurance and the detection gap in fraud prevention, live in the same place: the voice channel.

Voice Security

Voice Security closes the detection gap, and it treats trust as part of the customer experience rather than a separate security checkbox. A customer who cannot trust the voice on the line cannot stay loyal. Voice Security addresses the detection gap with three capabilities, each covering a distinct threat vector:

• Deepfake Detection analyzes inbound caller audio in real time and alerts agents when a synthetic or cloned voice is detected, before any account access, transaction, or credential reset takes place.
• Fraud Detection monitors the live call transcript for behavioral and linguistic signals of social engineering and delivers advisory alerts to agents while there is still time to act.
• Agent Voice Verification continuously confirms that the person on each agent call matches the enrolled agent's voice profile, detecting unauthorized substitution in remote and distributed operations.

All three run within Krisp’s Call Center AI platform, working across any CCaaS platform without the need for integrations or new infrastructure.

Speech Analytics

Speech Analytics addresses the coverage gap by automatically analyzing every call as soon as it ends. Conversations are scored against each organization’s own quality standards, flagged for potential compliance violations, summarized, and categorized, without manual review.

Violations surface post-call across every conversation, not weeks later in a scheduled audit. Supervisors get agent and team-level performance data across every call, making coaching cycles faster and driving continuous improvement. And call dispositions are generated automatically from the transcript, reducing after-call work and producing consistent categorization data across all agents.

"The voice channel carries a company's most sensitive moments, and for thirty years it ran on one assumption: if the voice sounds human, it is human. That assumption is now a liability," said Harry Folloder, Chief Commercial Officer at Krisp. "Voice Security answers it without taking the human out of the loop. The AI catches what an agent cannot hear, and the agent still makes the call. Speech Analytics and Voice Security are built for the scale and the threat environment contact centers are actually operating in right now."

"The voice channel is where customer relationships are made or broken," said Davit Baghdasaryan, CEO and Co-Founder of Krisp. "We built Krisp to transform the live conversation, not work around it. These two launches extend that into governance-every call is on record, every fraud signal is surfaced while there is still time to act, and every conversation is scored, not just the ones that happen to get reviewed."

Krisp Speech Analytics is generally available for contact centers and Voice Security is available in early access. Contact centers interested in joining the Voice Security early access program can request access here.

About Krisp

Founded in 2017, Krisp builds real-time Voice AI for meetings, contact centers, and developers. Its real-time voice AI technology enhances digital voice communication through bidirectional accent conversion, live speech-to-speech translation in over 60 languages, agent assist, voice security and speech analytics–all built on patented, award-winning noise cancellation models. Krisp applications are privacy-first, using industry-standard security practices across all audio hardware configurations and applications that support digital voice communication. Today, Krisp is deployed on over 200 million devices and processes over 80 billion minutes of voice conversations every month, with innovations recognized by Forbes, TIME, Gartner, and Fast Company, and won two 2026 Webby Awards for Technical Achievement.

Learn more about Krisp Call Center AI

Fonte: Business Wire