Block has introduced a voice AI model called Dessa, and the company’s own demos quickly set off two very different reactions: admiration for how human it sounds, and concern that tools like this could make phone scams and other forms of deception harder to spot.
Dessa is trained on speech, not text
According to Block, Dessa was trained entirely on human speech rather than text. The model can also listen while it is speaking, which lets it deal with interruptions, laughter and shifts in tone during a conversation.
Jack Dorsey wrote on X: 「Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence」.
In one demo, Dessa laughs in the middle of a sentence. In another, it responds to a surprise with what sounds like genuine emotion. Block said the conversations were recorded live and that no one was directing the model’s responses during the exchanges.
Direct audio processing and local operation on a MacBook Pro
Block said Dessa handles audio directly, instead of using the more common approach of turning speech into text first and then producing a spoken answer. The company also said it trained the model from scratch using a single Nvidia GB200 NVL72 computing system.
In Block’s description, the model is small enough to run locally on a MacBook Pro equipped with an M4 Max chip or better.
Dessa is not a public product at this stage. Block is presenting it as a research preview.
Social media response split between excitement and concern
The demos drew fast attention on social media, where some users focused on how natural the interactions sounded.
X user Alexander Hoff said he had never 「heard anything artificial that sounded so… human」. Another user, Hesamation, called it 「straight up sci-fi stuff」.
Binance founder Changpeng “CZ” Zhao also commented, pointing to possible education use cases. He said natural voice interaction could help children learn through conversation.
Zhao wrote: 「This might be very good for GiggleAcademy. Teaching kids in current ‘AI voice’ was a challenge」, referring to his education project.
Others focused on the risk side. Filmmaker Justine Bateman questioned why AI voices need to sound human at all and argued that the technology could be used to mislead people. She wrote: 「So you can trick people into lowering their guard」.
Journalist Shiv Aroor raised a similar warning about fraud, writing: 「Scam calls about to attain pandemic level」.
What Block has and has not disclosed
Dessa comes from Block, the company behind Square, Cash App, bitcoin wallet Bitkey and bitcoin mining hardware unit Proto.
Block’s AI research lab reports directly to Dorsey. It includes former staff from the 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. The audio samples available so far are company demonstrations, not independent tests.
The company is giving early access to a small group of AI safety researchers. Block said speech-based AI could eventually become better at recognizing emotional cues that text transcripts can miss, including distress and pain, while researchers examine risks the company may have overlooked.