# MIT Technology Review reports on AI video and speech watermarking

[Canonical reading](https://escapemeridian.com/readings/mit-technology-review-reports-on-ai-video-and-speech-0d656f98)

MIT Technology Review’s June 19, 2024, edition of The Download reports on a text-to-video model from Kuaishou and Meta’s system for watermarking AI-generated audio. The newsletter says Kuaishou released Kling for public testing, while OpenAI’s Sora remained inaccessible to the public at that point. It describes AudioSeal as a system that embeds hidden signals, or watermarks, in AI-generated audio and says it can pinpoint which parts of a longer recording may have been generated by AI. The newsletter says this could help detect AI-generated content and tackle misinformation and scams using voice cloning tools. It does not report how well AudioSeal performs in practice or whether it has reduced those harms. The edition also features a report on pneumatic tubes and a story about sonification, which it says could open access to education and careers for blind and visually impaired people.

Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.

## Original source

MIT Technology Review · Article

Original source date: Jun 19, 2024 (2024-06-19T12:10:00Z).

Date basis: Date from the publisher’s own records, through its API.

Added to Escape Meridian: 2026-11-18T12:23:13.061Z. First observed: 2026-10-29T02:04:03.268Z. Retrieved: 2026-10-29T02:05:33.855Z.

- [The Download: video-generating AI, and Meta’s voice cloning watermarks](https://www.technologyreview.com/2024/06/19/1094041/the-download-video-generating-ai-and-metas-voice-cloning-watermarks/)

## Limits

The newsletter gives no performance measurements or evaluation method for AudioSeal, and reports no measured effect on misinformation or scams. Text as retrieved on October 29, 2026; originally published June 19, 2024.


## Open question

How accurately can AudioSeal identify AI-generated portions of longer recordings in practice?


## Source-stated

MIT Technology Review’s newsletter reports that Kuaishou released Kling for public testing, while Sora remained inaccessible to the public at that point.

> Kling has already started letting people try the model themselves.

- [Supporting source](https://www.technologyreview.com/2024/06/19/1094041/the-download-video-generating-ai-and-metas-voice-cloning-watermarks/)

## Source-stated

MIT Technology Review reports that Meta created AudioSeal to embed watermarks in AI-generated audio; the newsletter says it can pinpoint AI-generated portions of longer recordings.

> The tool, called AudioSeal, is the first that can pinpoint which bits of audio in, for example, a full hour-long podcast might have been generated by AI.

- [Supporting source](https://www.technologyreview.com/2024/06/19/1094041/the-download-video-generating-ai-and-metas-voice-cloning-watermarks/)

## Source-stated

The newsletter says AudioSeal could help detect AI-generated content and address misinformation and scams using voice cloning tools.

> It could help to tackle the growing problem of misinformation and scams using voice cloning tools.

- [Supporting source](https://www.technologyreview.com/2024/06/19/1094041/the-download-video-generating-ai-and-metas-voice-cloning-watermarks/)

## Source-stated

MIT Technology Review says sonification could expand access for blind and visually impaired people.

> sonification could be transformative—opening access to education, to once unimaginable careers, and even to the secrets of the universe.

- [Supporting source](https://www.technologyreview.com/2024/06/19/1094041/the-download-video-generating-ai-and-metas-voice-cloning-watermarks/)

## Source and contributor context

MIT Technology Review · Publisher

The publisher is identified here. No individual author has been inferred.


## Source media

Media retrieval status: unchecked.


## Terms and references

Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.


## OpenAI

OpenAI develops AI models and products and publishes research, release announcements and evaluations.

- [Explore this context](/entities/openai)
- [Reference](https://openai.com/news/)

## Editorial provenance

Version 1. Publication ID: 2cd67e89-9048-e730-3095-736279b8ca4d.

Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.

Source check: passed by anthropic/claude-opus-5.5 on 2026-11-18T12:19:10.148Z. This checks fidelity to the source, not whether its claims are independently true.

Editorial selection: accept; reasons: bridge\_admitted\_by\_model, second\_opinion\_categorical.

Observation: 97a19e1418bc888b55d275e0ed3e1a90aec833a86f257705a7f16d8d0c562b1a. Source text hash: 805c52aba0e7b91b746f1b7d84a3d4dc91d27b3bf05efc7b85f7e90d4611130c.


## Revision and source-check history

### Version 1 — 2026-11-18T12:23:13.061Z

MIT Technology Review’s June 19, 2024, edition of The Download reports on a text-to-video model from Kuaishou and Meta’s system for watermarking AI-generated audio. The newsletter says Kuaishou released Kling for public testing, while OpenAI’s Sora remained inaccessible to the public at that point. It describes AudioSeal as a system that embeds hidden signals, or watermarks, in AI-generated audio and says it can pinpoint which parts of a longer recording may have been generated by AI. The newsletter says this could help detect AI-generated content and tackle misinformation and scams using voice cloning tools. It does not report how well AudioSeal performs in practice or whether it has reduced those harms. The edition also features a report on pneumatic tubes and a story about sonification, which it says could open access to education and careers for blind and visually impaired people.

## Original source

MIT Technology Review · Article

Original source date: Jun 19, 2024 (2024-06-19T12:10:00Z).

Date basis: Date from the publisher’s own records, through its API.

Added to Escape Meridian: 2026-11-18T12:23:13.061Z. First observed: 2026-10-29T02:04:03.268Z. Retrieved: 2026-10-29T02:05:33.855Z.

- [The Download: video-generating AI, and Meta’s voice cloning watermarks](https://www.technologyreview.com/2024/06/19/1094041/the-download-video-generating-ai-and-metas-voice-cloning-watermarks/)

## Limits

The newsletter gives no performance measurements or evaluation method for AudioSeal, and reports no measured effect on misinformation or scams. Text as retrieved on October 29, 2026; originally published June 19, 2024.


## Open question

How accurately can AudioSeal identify AI-generated portions of longer recordings in practice?


## Source-stated

MIT Technology Review’s newsletter reports that Kuaishou released Kling for public testing, while Sora remained inaccessible to the public at that point.

> Kling has already started letting people try the model themselves.

- [Supporting source](https://www.technologyreview.com/2024/06/19/1094041/the-download-video-generating-ai-and-metas-voice-cloning-watermarks/)

## Source-stated

MIT Technology Review reports that Meta created AudioSeal to embed watermarks in AI-generated audio; the newsletter says it can pinpoint AI-generated portions of longer recordings.

> The tool, called AudioSeal, is the first that can pinpoint which bits of audio in, for example, a full hour-long podcast might have been generated by AI.

- [Supporting source](https://www.technologyreview.com/2024/06/19/1094041/the-download-video-generating-ai-and-metas-voice-cloning-watermarks/)

## Source-stated

The newsletter says AudioSeal could help detect AI-generated content and address misinformation and scams using voice cloning tools.

> It could help to tackle the growing problem of misinformation and scams using voice cloning tools.

- [Supporting source](https://www.technologyreview.com/2024/06/19/1094041/the-download-video-generating-ai-and-metas-voice-cloning-watermarks/)

## Source-stated

MIT Technology Review says sonification could expand access for blind and visually impaired people.

> sonification could be transformative—opening access to education, to once unimaginable careers, and even to the secrets of the universe.

- [Supporting source](https://www.technologyreview.com/2024/06/19/1094041/the-download-video-generating-ai-and-metas-voice-cloning-watermarks/)

## Source and contributor context

MIT Technology Review · Publisher

The publisher is identified here. No individual author has been inferred.


## Source media

Media retrieval status: unchecked.


## Terms and references

Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.


## OpenAI

OpenAI develops AI models and products and publishes research, release announcements and evaluations.

- [Explore this context](/entities/openai)
- [Reference](https://openai.com/news/)

## Editorial provenance

Version 1. Publication ID: 2cd67e89-9048-e730-3095-736279b8ca4d.

Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.

Source check: passed by anthropic/claude-opus-5.5 on 2026-11-18T12:19:10.148Z. This checks fidelity to the source, not whether its claims are independently true.

Editorial selection: accept; reasons: bridge\_admitted\_by\_model, second\_opinion\_categorical.

Observation: 97a19e1418bc888b55d275e0ed3e1a90aec833a86f257705a7f16d8d0c562b1a. Source text hash: 805c52aba0e7b91b746f1b7d84a3d4dc91d27b3bf05efc7b85f7e90d4611130c.


### passed — 2026-11-18T12:19:10.148Z — anthropic/claude-opus-5.5

Publication: 2cd67e89-9048-e730-3095-736279b8ca4d

attribution — pass: The reading attributes each item to MIT Technology Review's newsletter The Download. The issue comments on several stories in its own words, so it counts as one piece rather than a bare listing. Claims about Kling, AudioSeal and sonification stay attributed as the newsletter's statements.

> This is today's edition of The Download, our weekday newsletter

summary — pass: The reading leads with the two AI items, Kuaishou's Kling and Meta's AudioSeal, and names their actors. It explains watermarks in plain language, notes the claimed use against voice-cloning scams and misinformation, and states what is unknown about performance. Sonification is peripheral and goes unexplained, which does not block understanding of the central development. The title and question fit the evidence.

> Meta has created a system that can embed hidden signals, known as watermarks, in AI-generated audio clips

date — pass: The reading's date, June 19, 2024, matches the UTC day of evidence.originalPublished (2024-06-19T12:10Z). The Deep Dive teasers describe possibly later events, but the reading neither uses nor relies on them. It frames Sora's inaccessibility as true at that point, with no hindsight. The source is a publisher page, not an archive copy, so no archive disclosure is required.

> Sora, which still remains inaccessible to the public four months after OpenAI debuted it

support — pass: Every claim is source-stated and matches its excerpt. The reading presents AudioSeal's potential benefits conditionally, as the newsletter's 'could help', and correctly notes that no performance data or measured effect is reported. The question follows from that gap without assuming an answer.

> It could help to tackle the growing problem of misinformation and scams using voice cloning tools.

media — pass: No verified media candidates were supplied, and the reading selects none, which is consistent.

## Recorded references

Explicit citations in public readings to this reading. Shared topics are not citations. Other record types need an explicit reference adapter.

No public references recorded here yet.

[Reference data (JSON)](/readings/0d656f984f72181f1ed0763cd5823ac3f6402fb40cb4fb9843a78b0b61bf775f/references.json)

## Corrections and challenges

[Send a correction or an opposing source](https://escapemeridian.com/corrections/)

[JSON record](https://escapemeridian.com/readings/mit-technology-review-reports-on-ai-video-and-speech-0d656f98.json)
