← Back to archive

Sesame’s voice AI feels human and raises scam concerns

Sesame recently released a demo of its voice AI. It doesn’t just generate speech; it feels human. It breathes, chuckles, stumbles over words, and even corrects itself. These little imperfections are so ... real.

A few highlights that stood out: 1️⃣ Realism beyond the uncanny valley; you can easily be emotionally attached to the AI voices, named Miles and Maya. Conversations with them aren’t robotic; they’re dynamic, expressive, and full of human-like nuances. 2️⃣ Sesame’s model integrates text and audio in a single-stage process rather than the traditional two-step text-to-speech pipeline. This gives it a much more natural flow. 3️⃣ Potential for deception. The ability to generate hyper-realistic AI voices could supercharge voice phishing scams. Some people have already started using “secret phrases” with family members to verify identities. A little scary, right?

It’s clear that conversational AI is evolving at an incredible pace, and Sesame is pushing the boundaries. Whether it’s for better or worse is up for debate, but one thing is certain: AI interactions are about to get way more personal.

Originally posted on X

PS: I'm pretty active on 𝕏 if you'd like to follow along

Archive

2026

2025

2024

2018