Why A Laughing AI Voice Doesn’t Tell You Everything

🔍 Read the full analysis: Why A Laughing AI Voice Doesn’t Tell You Everything on ThorstenMeyerAI.com

Before you orderOffer from Amazon

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

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

Block announced Dessa, a speech-to-speech research model that its demonstrations show laughing, interrupting and responding to emotional cues, but no outside researcher has tested the model. The company’s October 23 announcement has drawn attention both for its natural-sounding conversations and for the possibility that more convincing AI voices could make impersonation scams harder to spot.

Block says it trained Dessa from scratch on recorded human conversation alone, without text or pretrained models. The company describes the system as full-duplex: it can listen while speaking, respond to pauses and interruptions, and produce cues such as laughter or a sigh. Block also says it runs faster than real time on a MacBook Pro M4 Max or better, but has not published a latency figure.

The public evidence consists of recordings selected by Block. The company says each clip is a live conversation and that nobody steered the exchanges. The Verge reported that the chats could sound more natural than a competing system but were sometimes awkward or uncanny. TechCrunch said it was unclear whether emotion in the demonstrations was guided by tags.

As of October 26, Block had published no technical paper, model card, benchmark results, weights or API, and had not announced a public release date. Access was described as a waitlist and early access for a small group of AI safety researchers. Block’s project page also leaves open how well the model learns other languages.

Natural Speech Raises Fraud Concerns

The practical concern is not that Dessa has been shown to commit fraud; there is no evidence in the source material that it has. Rather, lifelike timing, laughter and responses to interruptions could make synthetic voices more persuasive if similar tools are used to impersonate someone. The demonstrations do not establish that Dessa can copy a specific person, but they illustrate why a voice alone may be a weak way to verify identity.

The FBI’s 2025 Internet Crime Report, published in April 2026, counted 22,364 complaints mentioning AI, with $893 million in reported losses. It also reported more than $5 million in losses to distress scams using cloned voices to imitate loved ones. Those figures describe reported complaints and losses across AI-related fraud; they do not show that Dessa caused any of them or measure the effect of this model.

For businesses, the development is a reason to avoid treating a familiar voice as sufficient authorization for money transfers or account changes. Independent callbacks to known numbers and additional verification can reduce reliance on voice recognition. For families, a private code word and calling back through a saved number are practical safeguards when a caller demands urgent help or money.

What Block Has Disclosed So Far

Block, the company behind Square and Cash App, published its Dessa project page on October 23, 2026. Co-founder Jack Dorsey described talking with the model as feeling like talking with a person, with emotion, nuance, pauses and natural cadence. Dessa lead Ragavan Thurairatnam called the work “a GPT3 moment for speech.” These are company and project-team characterizations, not independent findings.

Block’s demonstrations include laughter during a sentence, a response to someone struggling to recall a movie title, and an exchange about an anniversary surprise. The company says the model chooses its own voice and personality. Coverage by The Verge, TechCrunch and CoinDesk has emphasized that the recordings cannot substitute for access to the model, published methods or independent benchmarks.

The safety debate is not limited to fraud. Block argues that a voice system able to detect emotion could respond more appropriately when a person is distressed. Critics have questioned why AI voices need to sound human. Both arguments concern the same capability: more expressive speech may improve interaction while making synthetic voices harder to distinguish.

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

— Jack Dorsey, Block co-founder

Capabilities and Safeguards Remain Unverified

The central limitation is that outsiders have not run Dessa. Without a paper, model card, benchmarks or public access, it is not possible to independently assess its performance, limitations, training details or reliability beyond Block’s chosen examples. The company has not disclosed the model’s size or a public release date.

Other open questions include which languages it supports, whether its emotional responses are guided by tags, and whether it can imitate a particular person. The source material also identifies no published details about watermarking, consent checks or other safeguards. These gaps do not prove that protections are absent; they mean their design and effectiveness have not been established publicly.

The demonstrations show speech that can sound expressive. They do not establish that the model experiences emotion, understands a conversation as a person would, or performs consistently outside those clips. Nor does the available reporting show that Dessa has been used in a scam.

Release Details Will Shape Assessment

The next meaningful step would be public technical documentation or access for independent researchers. That could allow testing of performance, language coverage, consent safeguards and misuse risks beyond the recordings Block selected. As of October 26, the company had not provided a release date, so it is unclear when broader testing might happen.

Until then, organizations can review procedures that authorize payments or account changes by voice, while families can agree on a code word and verify urgent requests through a known number. These precautions address voice impersonation generally; they are not evidence of a specific threat from Dessa.

Key Questions

What is 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 and speak at the same time.

Can the public use or test Dessa?

Not broadly, based on information available on October 26. Block described a waitlist and early access for a small group of AI safety researchers, but announced no public release date, weights or API.

Has independent testing confirmed the model’s abilities?

No. Public assessments are based on Block’s selected recordings and journalists’ listening impressions. No outside researcher has run the model, and Block has not published benchmarks or a technical paper.

Does the announcement show that Dessa can clone a specific person?

No. The available material does not establish that capability. Whether Dessa can imitate a specific person remains unclear, as do details about consent checks and watermarking.

How can people respond to urgent voice requests for money?

Do not rely on a familiar-sounding voice alone. Hang up and call a number already known to you; families can also agree on a private code word. These are general precautions against impersonation scams, not a response to a confirmed Dessa-related incident.

Source: ThorstenMeyerAI.com