The Onion’s Scarlett Johansson OpenAI Satire Explained

Updated 5 November 2026 · Standard formula, shown below

The Onion’s Scarlett Johansson–OpenAI satire hit the mark because it condensed a real, widely discussed dispute into a familiar comic premise: a company’s AI voice sounded strikingly like a famous actor’s, despite her saying she had declined an invitation to voice it. The joke worked as satire not because the disputed voice was proved to be Johansson’s, but because the episode exposed tensions around consent, resemblance, corporate power and the marketing of human-like AI.

The dispute behind the joke

In May 2024, OpenAI introduced GPT-4o, a model designed to handle text, images and audio more naturally. One of the voices available in its ChatGPT voice experience was called Sky. Listeners quickly remarked that Sky sounded like Johansson, especially to people who remembered her performance as Samantha, an AI operating system, in the 2013 film Her.

Johansson said OpenAI CEO Sam Altman had contacted her in September 2023 to ask whether she would voice ChatGPT. She said she declined after careful consideration. According to her statement, Altman contacted her agent again shortly before the GPT-4o announcement, asking her to reconsider; she said she declined again. When Sky was released, she and people close to her were surprised by its resemblance to her voice.

OpenAI said Sky was not an imitation of Johansson and had been voiced by a different professional actor. The company said it had selected the voice through a casting process and had not intended it to resemble her. It paused Sky’s use after the controversy. Those distinctions matter: a perceived resemblance is not, on its own, proof that a voice was copied or that a particular person’s recording was used.

Why the satire landed

The Onion’s premise connected the news to a recognizable comic pattern: a powerful organization says it is offering a convenient, futuristic service while the public notices an uncomfortable human cost. A voice assistant is supposed to feel helpful and personal. If that voice seems to belong to a well-known person who did not agree to the association, the polished product begins to look like a question about who gets to use someone’s identity.

That contrast gives the episode its satirical force. The technology was presented as a major leap toward natural conversation, but the public discussion became less about technical capability than about a basic social expectation: people should have a say in whether their voice is used to represent a product. The story made an abstract debate about synthetic media concrete and easy to grasp.

Recognition: Many people could immediately understand why a familiar voice might feel personal.
Consent: Johansson’s account of declining the offer gave the story a clear conflict.
Power imbalance: A global technology company could launch a product at enormous scale, while the person allegedly evoked had to respond after the public heard it.
Irony: A fictional film about an intimate AI voice gave listeners an especially vivid reference point for a real AI voice product.

How the film connection shaped the reaction

Johansson’s role in Her was not a minor coincidence. She voiced Samantha, an AI character whose warmth and conversational style were central to the film. The GPT-4o launch presentation also emphasized natural, responsive conversation. That overlap made comparisons feel culturally loaded: listeners were not merely comparing vocal pitch or accent, but recalling a performance that had already shaped popular ideas about what an AI companion might sound like.

Altman’s post announcing GPT-4o included the word “her,” which many readers took as a pointed reference to the film. That detail amplified the sense of irony, although it does not establish that Sky was designed to imitate Johansson. The public could reasonably notice the cultural reference and the resemblance while still distinguishing those impressions from evidence about how the voice was created.

What the dispute does—and does not—prove

The episode is often summarized as “OpenAI copied Scarlett Johansson’s voice.” That wording goes beyond what was publicly established at the time. Johansson said the voice sounded like her and described her prior refusals; OpenAI denied that Sky was an imitation and said a different actor had performed it. The existence of a resemblance, the identity of the performer and the intent behind the design are separate questions.

It is also useful to distinguish a voice that sounds similar from one generated by cloning a person’s recordings. Modern voice systems can produce speech from a selected or trained voice, but a listener cannot determine the production method simply by hearing a clip. Public statements, documentation and evidence about the process would be needed to make stronger claims. The satire could criticize the surrounding power dynamics without settling those technical or factual questions.

Why it mattered beyond one voice

The controversy arrived as generative AI products were becoming more visible and more capable of producing convincing audio. That raised practical questions for actors, voice performers and the wider public: what counts as an acceptable resemblance, how permission should be obtained, and how clearly companies should explain who or what is speaking. A voice can be recognizable even when it is not an exact copy, and familiar vocal traits can carry commercial and emotional value.

It also showed how quickly product design becomes a question of trust. Users may welcome natural speech but still want to know whether a voice belongs to a real performer, whether that performer consented, and what happens if a voice is later changed or withdrawn. OpenAI’s decision to pause Sky did not resolve every question, but it demonstrated that public response can affect a product’s availability.

For satire, the episode offered a compact target: the gap between a company’s claims of innovation and the ordinary expectations of respect and permission. The joke did not need to prove that a crime or deliberate impersonation had occurred. Its target was the uneasy possibility that a voice could feel unmistakably personal while the systems and decisions behind it remained opaque.

Frequently Asked Questions

Did OpenAI use Scarlett Johansson’s actual voice?
There was no public confirmation that Sky used Johansson’s recordings. Johansson said it sounded like her, while OpenAI said Sky was voiced by a different actor and was not intended to imitate her.

Why did people compare Sky to the voice in Her?
Johansson played the AI Samantha in the film, and listeners heard a resemblance between that familiar performance and Sky. The GPT-4o launch’s emphasis on conversational AI made the comparison especially immediate.

Did Scarlett Johansson agree to voice ChatGPT?
Johansson said she declined OpenAI’s request in 2023 and declined a later approach shortly before the GPT-4o announcement. OpenAI did not use her as the performer for Sky, according to the company.

What happened to the Sky voice?
OpenAI paused Sky’s use after the controversy. The company said the pause was not intended to imply that the voice was an imitation of Johansson.

Why did The Onion’s satire resonate?
It drew on a real dispute that combined celebrity, AI, consent and corporate influence. The premise was easy to recognize, while the gap between a human-sounding product and uncertainty about permission supplied the irony.

Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.