🔍 Read the full analysis: Block’s Dessa Interrupts Like A Person. How Much Should We Read Into It? on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Block announced Dessa, a speech-to-speech research model that it says can listen and speak at once, respond with emotional cues and run locally on high-end Mac hardware. The demonstrations are company-selected; no paper, benchmark, public model or independent test is available, so its performance and implications remain uncertain.
Block has presented Dessa, a research model that it says can hold spoken conversations with interruptions, laughter and other emotional cues, but outsiders cannot yet test the system. The demonstrations have prompted interest in more natural voice interfaces and concern about impersonation scams; there is no public model, technical paper or independent benchmark to establish how reliably it works.
Block published a project page for Dessa on October 23. The company says the model was trained from scratch on recorded human conversation, without text or pretrained models, and can listen while it speaks. In its description, Dessa recognizes conversational signals such as pauses and backchannels and can respond with laughter or other vocal expressions. Block also says it runs faster than real time on a MacBook Pro M4 Max or better, though it has not provided a latency figure.
The examples on Block’s page include a model laughing mid-sentence, helping a speaker recall a film title and reacting to an anniversary surprise. Block describes the recordings as live conversations that were not steered. Those are still company-selected demonstrations, rather than tests conducted by outside researchers. The Verge reported that some exchanges sounded awkward or uncanny, while finding Dessa more natural than another model in a comparison. TechCrunch noted that it was unclear whether emotional delivery was shaped by tags.
Block co-founder Jack Dorsey wrote that talking with Dessa feels like talking with a person, citing its emotion, nuance, pauses and cadence. Dessa lead Ragavan Thurairatnam called it a “GPT3 moment for speech” and said it could run on consumer GPUs. Those remarks describe the company’s view of the system; they do not substitute for public performance results. Block has not disclosed the model’s size, benchmark scores or a release date, according to CoinDesk.
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
Natural Voices Raise Fraud Risks
Dessa matters partly because a more fluid voice interface could make speech systems easier to use. Block argues that current voice AI often misses emotional cues and can fail to respond appropriately when a person is distressed. A system that recognizes pauses, interruptions or tone could support more responsive conversations, although the demonstrations do not establish how well it would perform in practical settings.
The same qualities can make a synthetic voice harder to distinguish from a person. Voice impersonation scams already exist; they do not depend on Dessa. The FBI’s 2025 Internet Crime Report, published in April 2026, counted 22,364 complaints mentioning AI and reported $893 million in losses associated with those complaints. The report also recorded more than $5 million in losses to distress scams using cloned voices of loved ones. These figures describe reported cases and losses, not proof that Dessa or any Block product has been used in fraud.
For readers, the practical point is that a familiar-sounding voice should not be treated as proof of identity, particularly when a caller demands urgent money or sensitive information. Businesses and families can use verification steps independent of the voice, such as calling back on a known number or agreeing on a family code word. These are general precautions, not evidence that a particular scam wave has resulted from Dessa.
As an affiliate, we earn on qualifying purchases.
A Demo, Not a Public Release
Dessa comes from a small research lab at Block, the company behind Square and Cash App. CoinDesk reported that the team was formed by former members of Toronto deep-learning startup Dessa, which Square acquired in 2020. Block says the model learned from “hundreds of times fewer words” than large language models require and was trained using one NVIDIA GB200 NVL72. The company has not published enough technical detail for outside parties to verify those claims or compare the system with other models.
As of October 26, Block’s page offered a waitlist and early access for a small group of AI safety researchers. The company has announced no downloadable weights or API. That distinction matters: viewers can hear selected examples, but researchers, companies and journalists cannot independently examine the model’s behavior across varied speakers, languages or stressful situations.
Public reaction has ranged from praise for the conversational quality to concern that human-sounding voices could lower listeners’ guard. Block has framed emotional awareness as a potential benefit, while critics have questioned why AI voices should sound human. The debate reflects competing uses of the same capability; it does not establish that the model can convincingly imitate a particular person.
“Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.”
— Block co-founder Jack Dorsey
Performance and Safeguards Remain Unknown
There is no public technical paper, model card, benchmark, license or general release date. The model’s size, language coverage, error rates and performance across different accents and conversational situations have not been disclosed. Block’s demonstrations cannot establish how consistently Dessa handles interruptions or emotion outside the clips selected for its page.
It is also unclear whether the system can imitate a specific person, what consent checks may apply, or whether its output will be watermarked or otherwise identifiable. Block says Dessa selects its own voice and personality, but that does not answer how the system might be used or controlled. The company has not announced that Dessa has been used in a scam, and no such link is established by the source material.
The FBI figures provide a measure of reported AI-related complaints and losses in 2025, but they do not isolate losses caused by voice models like Dessa. Nor do the available reports show how often victims can distinguish a clone from a real person. Those limits make it premature to quantify any added fraud risk from this particular model.
Watch for Testing and Release Details
The next useful evidence would be independent access to the model, published technical documentation and evaluations of its conversational performance and safety controls. Block has said a small group of AI safety researchers will receive early access, but it has not named a public release date or detailed what that testing will cover.
Until those details emerge, readers should treat the recordings as demonstrations of what Block says its research model can do, not as a full account of real-world reliability. For any unexpected request involving money or private information, verify the caller through a separate, previously trusted channel rather than relying on the voice alone.
Key Questions
What is Block’s Dessa?
Dessa is a speech-to-speech research model that Block says can listen and speak at the same time and respond to conversational cues. Block says it was trained from scratch on human conversation, without text or pretrained models.
Can the public try or download Dessa?
Not as of October 26, 2026. Block has offered a waitlist and early access for a small group of AI safety researchers, but has announced no public model weights or API.
Has Dessa been shown to imitate a specific person?
The available information does not establish that. Block says the model selects its own voice and personality, but has not clarified whether it can imitate a specific individual or what consent checks would apply.
Does the FBI data show that Dessa is being used in scams?
No. The FBI’s 2025 report counted 22,364 complaints mentioning AI and reported losses, including more than $5 million in distress scams using cloned voices. Those figures do not show that Dessa was involved.
How can people verify a suspicious call?
Do not rely on a familiar-sounding voice alone. Hang up and call the person back using a number you already trust, or use a pre-agreed family code word; treat pressure to act urgently as a reason to verify independently.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
