Responses are divided between awe at its human-like quality and concerns it could power scams and deception, while CZ sees potential for children’s education.

• Block introduced Dessa, a voice AI trained exclusively on human speech that speaks without text and operates locally on a MacBook Pro with an M4 Max chip.
• Reactions are split between wonder at its human-like sound and worries it could enable scams and deception, while CZ eyes applications in children’s education.
• Block has not revealed the model’s size, benchmarks, or a release date; the research preview is initially going to a small group of AI safety researchers.

Jack Dorsey’s payments firm Block XYZ$76.46·Market Closed has revealed a voice AI model that sounds strikingly human in company demos, eliciting responses from awe at its conversational skills to warnings that the tech could make phone scams more difficult to spot.

Named Dessa, the model was trained solely on human speech instead of text and can listen while speaking, enabling it to manage interruptions, laughter, and tone shifts in conversation, per Block’s announcement.

“Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence,” Dorsey wrote on X.

In one demo, Dessa laughs mid-sentence; in another, it reacts to a surprise with what seems like genuine emotion. Block states the conversations were recorded live with no one directing the model’s replies.

Unlike systems that convert speech to text before generating a spoken reply, Dessa processes audio directly. Block said it trained Dessa from scratch using a single Nvidia GB200 NVL72 system. The model is compact enough to run locally on a MacBook Pro with an M4 Max chip or better, according to the company.

For now, Dessa remains a research preview, not a publicly available product.

Awe and Alarm

Dessa’s demo quickly garnered attention on social media, with some users praising how natural the interactions sounded.

X user Alexander Hoff said he had never “heard anything artificial that sounded so… human,” while user Hesamation called it “straight up sci-fi stuff.”

Binance founder Changpeng “CZ” Zhao also weighed in, highlighting potential educational uses where natural voice interactions could help children learn through conversation.

“This might be very good for GiggleAcademy. Teaching kids in current ‘AI voice’ was a challenge,” Zhao wrote, referring to his education project.

However, the same humanlike qualities that impressed some listeners raised concerns among others about the technology’s potential for deception.

Filmmaker Justine Bateman questioned why AI voices need to sound human, arguing it could be used to mislead people. “So you can trick people into lowering their guard,” she wrote.

Journalist Shiv Aroor voiced similar fraud concerns, writing, “Scam calls about to attain pandemic level.”

Dessa originates from Block, which runs Square, Cash App, bitcoin wallet Bitkey, and bitcoin mining hardware unit Proto. Its AI research lab reports directly to Dorsey and includes former staff from Toronto-based AI startup Dessa, which Square acquired in 2020.

Block has not disclosed the model’s size, benchmark results, or a public release date. Its audio samples remain company demos, not independent tests.

The company is offering early access to a small group of AI safety researchers. Block says speech-based AI could eventually excel at recognizing emotional cues like distress and pain that text transcripts miss, while researchers evaluate risks the company may have overlooked.