🔍 Read the full analysis: Why Listening To AI Is No Longer Enough To Know What You’re Hearing 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 trained on recorded human conversation, and published demonstrations of it laughing, interrupting and responding with apparent emotion. No independent researcher has publicly tested the model, and Block has released no paper, benchmarks, weights or API. Its claims remain unverified, while existing voice-fraud losses show why human-sounding calls should not be treated as proof of identity.
Block announced Dessa, a speech-to-speech research model that the company says was trained from scratch on recorded human conversation, and shared demonstrations of it laughing, interrupting and reacting with apparent emotion. The announcement matters because a more natural-sounding voice could make synthetic calls harder to recognize, but no one outside Block has publicly tested the model, and the company has not released its technical documentation or model.
Block, the company behind Square and Cash App, posted a project page for Dessa on October 23. It says the model uses conversation audio rather than text or pretrained models, and can listen while speaking, respond to interruptions and produce features such as laughter or a sigh. Block also says it runs faster than real time on a MacBook Pro M4 Max or better. It has not published a latency measurement.
The public evidence so far consists of recordings selected by Block. In them, Dessa laughs mid-sentence, helps a speaker recall a film title and responds to an anniversary surprise with what the source describes as teary-sounding gratitude. Block says each is a live recording and that nobody steered the model. Those statements have not been independently verified.
Outside coverage has noted both the model’s natural delivery and its awkward moments. The Verge described some exchanges as “a little awkward and uncanny”; TechCrunch said it was unclear whether emotion might be guided by tags. As of October 26, there is no public paper, model card, benchmark, license, API or model weights. Block has offered a waitlist and early access to a small group of AI safety researchers, but no general release date.
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
Why Natural Voices Affect Fraud Checks
A voice that sounds relaxed, emotional or responsive can make a call feel personal. That quality may be useful in customer service, but it is also relevant to impersonation scams, where the caller’s apparent identity and urgency can prompt someone to send money or disclose information. Dessa has not been shown to be used in a scam; the concern is that more convincing synthetic speech could reduce the usefulness of familiar cues such as robotic timing or flat delivery.
The FBI’s 2025 Internet Crime Report, published in April 2026, counted 22,364 complaints mentioning AI and $893 million in reported losses. The FBI also said victims lost more than $5 million to distress scams using cloned voices to mimic a loved one. These figures cover reported incidents and do not establish that Dessa contributed to them. They show that voice cloning-related fraud predates Block’s announcement.
For organizations, the development is a reason not to treat a familiar-sounding voice as sufficient authorization for payments or account changes. A separately verified callback and written approval can provide another check. For families, a pre-agreed code word and a return call to a known number can help test an urgent request. Those precautions address voice impersonation generally; they are not specific to Dessa.
Products Worth Considering
Bose QuietComfort Wireless Noise-Cancelling Over-Ear Headphones, Black
Sony WH-1000XM4C Wireless Noise-Cancelling Headphones, Black
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Black
As an affiliate, we earn on qualifying purchases.
Block’s Claims and Public Evidence
Block describes Dessa as a project from a small AI research lab formed by former members of Toronto-based deep-learning company Dessa, which Square acquired in 2020, according to CoinDesk. The company says the model was trained using a single NVIDIA GB200 NVL72 and learned from “hundreds of times fewer words” than large language models need. Those are company claims; Block has not published enough technical detail for independent evaluation.
Co-founder Jack Dorsey wrote that talking with Dessa feels like talking with a person, citing emotion, nuance, pauses and cadence. Project lead Ragavan Thurairatnam called it “a GPT3 moment for speech,” and said it could run on consumer GPUs. The comments describe the team’s view of the project, not results established through independent testing.
Block says Dessa chooses its own voice and personality. The company has not publicly answered questions about supported languages, consent checks, watermarking or whether the model can imitate a particular person. It has also described learning new languages efficiently as an open question. Those gaps limit what can be concluded from the English-language demonstrations currently available.
“Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.”
— Jack Dorsey, Block co-founder, in a post about Dessa
What Independent Tests Have Not Established
Without public access, technical documentation or independent testing, it is not possible to verify how consistently Dessa performs, how it handles different speakers and conditions, or how its apparent emotional expression is generated. Block has not disclosed the model’s size or benchmark results, and no public release date is known.
It is also unclear which languages it supports, what safeguards apply to voice imitation, whether consent is checked, and whether the audio carries a detectable mark. The demonstrations do not establish that the model can reliably impersonate a particular person, nor that it has been used in fraud. Reports about wider AI-related losses provide context, not evidence of a link to Dessa.
Access, Documentation and Safeguards
The next meaningful evidence would come from independent researchers testing the model and from Block publishing details on its design, performance, limitations and safeguards. The company has said a small group of AI safety researchers will receive early access, but it has not identified a public timetable for wider access or a technical release.
Until those details emerge, businesses should verify high-stakes requests through a separate channel rather than relying on a caller’s voice, and people should pause when a caller demands urgent action. Block’s announcement establishes that it has built and demonstrated a research system; how broadly it works and what protections accompany it remain open questions.
Key Questions
What is Block’s Dessa model?
Dessa is a speech-to-speech research model that Block says was trained from scratch on recorded human conversation. The company says it can listen and speak at the same time and respond with features such as laughter and pauses.
Can the public use or independently test Dessa?
As of October 26, 2026, Block has not released model weights, an API or a public technical report. It has described a waitlist and early access for a small group of AI safety researchers, but has not announced a general release date.
Has Dessa been linked to a voice scam?
No such link is established in the available material. The FBI has reported losses from voice-cloning distress scams, but those figures do not show that Dessa was involved.
How can someone check an urgent call asking for money?
Hang up and call the person back using a number you already know. Families can agree on a private code word; organizations can require approval through a separate, verified channel for payments or account changes.
Source: ThorstenMeyerAI.com
