🔍 Read the full analysis: Block’s Dessa Laughs, Hesitates And Interrupts Like A Person. Why A Human-Sounding Voice Is No Longer Proof Of Anything on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Block announced Dessa, a speech-to-speech research model whose company-selected demos feature laughter, interruptions and emotional cues. The model has not been publicly released or independently tested, so its performance and safeguards remain unverified. Its humanlike conversational style renews concerns about voice impersonation scams.
Block announced Dessa, a speech-to-speech research model, on October 23, showing recordings in which it laughs mid-sentence, responds to conversational cues and interrupts naturally. The company says it trained the model on human conversation alone, but the demonstrations are Block-selected and the system has not been made available for independent testing, leaving its capabilities and safeguards unverified.
Block, the company behind Square and Cash App, says Dessa was trained from scratch using human conversation without text or pretrained models. The company describes it as a full-duplex system: it can listen while speaking, respond to pauses and backchannel during a conversation rather than waiting for a complete spoken prompt. Its project page also says it can reproduce cues such as laughter, sarcasm and sighs.
Block says the model runs faster than real time on a MacBook Pro M4 Max or better, and that training used one NVIDIA GB200 NVL72. It has not provided a latency figure, model size or benchmark results. The company says access is limited to a waitlist and a small group of AI safety researchers; it has announced no public weights, API or release date.
The available evidence consists of recordings published by Block. The company says each is a live conversation and that nobody steered the model. The Verge, after listening, described some exchanges as occasionally awkward and uncanny, while finding Dessa more natural than another model in the comparison. TechCrunch questioned whether emotion might be guided by tags. These reports describe impressions of the demonstrations, not independent evaluations of the underlying system.
Infographic 01 · Dessa fact sheet
What Block claims, what was heard, what is missing.
Every capability below comes from Block or its selected recordings. Nobody outside Block has run the model yet.
| Topic | What we know | Status |
|---|---|---|
| Training data | Human conversation only: “no text, no pretrained models” | Block’s claim |
| Conversation | Full-duplex: listens while it talks, backchannels, cuts in | Block’s claim |
| Emotion | Hears and speaks laughter, sarcasm, sighs; unclear if steered by tags (TechCrunch) | Block’s claim |
| Speed | “Faster than real time”; no latency figure | Block’s claim |
| Hardware | Trained on one NVIDIA GB200 NVL72; runs on a MacBook Pro M4 Max or better | Block’s claim |
| Demos | Some chats “a little awkward and uncanny”, more natural than GPT-Live-1 (The Verge) | reporters heard it |
| Size, benchmarks, release date | Not disclosed (CoinDesk) | not disclosed |
| Paper, model card, license, weights | None found as of 26 Oct 2026 | not disclosed |
| Languages | Not listed; Block calls learning new languages an open question | not disclosed |
| Voice cloning safeguards | No statement on watermarking, consent checks or imitating a specific person | not disclosed |
Who can use it
research preview
- Public: a waitlist; “launch links are coming soon”
- Early access: “a small group of AI safety researchers”, institutional email required
- Business use: none announced; Square already uses voice AI for ordering (TechCrunch)
Who built it
from Block’s page
- A “(very) small AI research lab at Block” reporting to Jack Dorsey
- Founded by former members of Dessa, the Toronto startup Square acquired in 2020 (CoinDesk)
- Lead Ragavan Thurairatnam: “a GPT3 moment for speech”
Sources: Block, Dessa project page (23 Oct 2026); Ragavan Thurairatnam on X (23 Oct 2026); The Verge (23 Oct 2026); TechCrunch (24 Oct 2026); CoinDesk (24 Oct 2026).
Powered by Thorsten Meyer AI
When a Voice Sounds Convincing
Dessa’s presentation matters because the conversational cues it displays—laughter, pauses and interruptions—can make a voice interaction feel less scripted. That may help voice assistants respond more comfortably, but it also sharpens concerns about impersonation. The demonstrations do not establish that Dessa can clone a particular person or that it has been used in fraud. They do show why listeners may find it harder to judge a voice call by conversational style alone.
The source article cites FBI figures from its 2025 Internet Crime Report, published in April 2026: 22,364 complaints mentioned AI, with $893 million in reported losses. It also reports more than $5 million in losses from distress scams involving cloned voices of loved ones. These are figures about reported complaints and losses, not evidence that Dessa caused or contributed to them. Voice impersonation risks already exist; a new model’s potential effect on them remains unmeasured.
For organizations, a familiar-sounding voice should not be treated as sufficient proof of identity, especially when a caller requests payment or access. The source recommends verifying sensitive requests through a known number and requiring written consent for synthetic voices. For families, an agreed code word and a callback to a saved number offer practical checks. These precautions address voice fraud generally; no evidence presented shows that Dessa itself is being used to make scam calls.
Products Worth Considering
Sony WH-1000XM4C Wireless Noise-Cancelling Headphones, Black
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
Bose QuietComfort Headphones (2nd Gen), Wireless Noise-Cancelling Over-Ear Headphones, Eucalyptus Green
As an affiliate, we earn on qualifying purchases.
What Block Has Released So Far
Block published Dessa’s project page on October 23. Jack Dorsey, Block’s co-founder, wrote that speaking with it felt like talking with a person, citing emotion, nuance, pauses and natural cadence. Dessa lead Ragavan Thurairatnam called it “a GPT-3 moment for speech” and said it could run on consumer GPUs. Those descriptions are statements from Block figures, not independent findings.
The demonstrations include laughter during speech, a conversation in which the model helps recall a film title, and a response to an anniversary surprise. Block says Dessa chooses its own voice and personality. The company has not published a technical paper, model card, license or detailed account of how it evaluates the system. The Toronto deep-learning startup Dessa was acquired by Square in 2020, according to CoinDesk; Block’s current research team is described as a small lab that includes former members of that company.
Reporting around the announcement reflects both interest and concern. The Decoder cautioned that sounding emotional does not establish emotion or intelligence. Filmmaker Justine Bateman questioned why AI voices should sound human, while Block has argued that emotional awareness could help voice systems respond more appropriately to people. The competing arguments concern the same feature: greater naturalness may improve interaction, while making a synthetic voice harder to recognize.
“Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.”
— Jack Dorsey, Block co-founder
Independent Tests and Safeguards
No outside group has run the model, according to the source material available as of October 26. Without public access, technical documentation or benchmarks, it is not possible to verify Block’s claims about training, speed, conversational performance or hardware requirements. Company-selected recordings can show examples, but cannot establish how consistently the model performs across ordinary use.
Block has not stated which languages Dessa supports or explained whether its emotional responses are directed by tags. The source also reports no public information on watermarking, consent checks or protections against imitating a specific person. Block says the model chooses its own voice and personality, but that statement does not resolve whether it could reproduce an identifiable voice under some conditions. There is also no announced public release date.
The FBI statistics cited above describe reported AI-related complaints and losses across 2025, not the effect of Dessa. It remains unknown whether a model with the demonstrated style would change fraud rates, how reliably listeners could identify it as synthetic, or what safeguards would accompany wider access.
Access, Documentation and Testing
The immediate next step is any further release from Block: documentation, details about access for safety researchers, or a public timetable. Independent evaluation would be needed to test the company’s claims under repeatable conditions, including latency, conversational turn-taking, language coverage and performance outside curated demonstrations. Information on voice consent, identity safeguards and synthetic-audio labeling would clarify how the company intends to manage misuse risks.
Until that information appears, readers should treat Dessa as an announced research model rather than a widely available product. People and businesses can apply existing verification practices now: confirm high-stakes requests through a separate, trusted channel, and do not rely on a familiar-sounding voice alone. The source also points to EU AI Act Article 50, which it says has applied since August 2, 2026; the article’s specific requirements and how they apply to any particular deployment are not detailed in the material provided.
Key Questions
What is Block’s Dessa?
Dessa is a speech-to-speech research model announced by Block on October 23, 2026. Block says it was trained from scratch on human conversation and can listen while it speaks.
Can the public use Dessa?
Not based on information available as of October 26. Block has described a waitlist and early access for a small group of AI safety researchers, but announced no public API, model weights or release date.
Has Dessa been independently tested?
No independent model testing is reported in the source material. Journalists listened to Block’s recordings, but that is not the same as running the system or verifying its technical claims.
Does the announcement show that Dessa can clone a person’s voice?
No. The material says Block has not explained whether Dessa can imitate a specific person. The demonstrations show conversational speech, not verified voice cloning.
What can people do about voice impersonation scams?
For urgent requests involving money or sensitive information, hang up and call a number already known to you. Families can agree on a private code word. These steps address voice scams generally and are not evidence of Dessa-related fraud.
Source: ThorstenMeyerAI.com
