目錄
In June 2026, Los Angeles rapper Fenix Flexin and producer Purps on the Beat released the single “Rubberz.” The track debuted at No. 66 on the Billboard Hot 100 in its first week and continued to climb afterward. As of August 2, 2026, it had accumulated approximately 10.8 million streams on Spotify. The controversy was not caused solely by Fenix Flexin’s sudden change in musical style, but by a question that still has no definitive answer: was generative AI used to create the song?
The controversy shows that music suspected of using generative AI is already capable of entering mainstream charts, while the current industry still lacks a reliable system for determining exactly how much AI was involved. Audio-detection tools, live performances, production-session footage and statements from the creators all point in different directions. None of these forms of evidence is sufficient on its own to settle the matter. The information in this article is current as of August 2, 2026, and the story is still developing.
How Did the AI Controversy Around “Rubberz” Begin?
“Rubberz” was released on June 5, 2026. The song draws heavily from 1980s British synth-pop and new wave. Fenix Flexin performs with a British-style vocal delivery reminiscent of Morrissey, while the lyrics combine heartbreak, money and conspicuous consumption. The result is noticeably different from the West Coast rap music he previously released as a member of Shoreline Mafia.
A sudden change in musical style does not prove that AI was used. However, the discussion quickly focused on several specific details. Listeners reported unnatural digital artifacts in the drums and reverb, backing vocals that occasionally disappeared without a natural transition, and lyrical stresses that did not resemble typical human vocal phrasing. Music critics and industry figures including Anthony Fantano, Curtiss King and Charlie Harding publicly raised questions about the recording.
Why Did Treblo Become the Main Focus of the Controversy?
Treblo is a free AI music-generation tool formerly known as Sonauto. Users can enter the names of artists in prompts or style tags to generate songs with similar vocal and stylistic characteristics. This capability has also created legal concerns involving voice imitation and the authorization of training data.
Zinstrel and Stems Media both tested Fenix Flexin-related tags using Treblo. The resulting songs reportedly had vocal delivery and tonal characteristics that closely resembled “Rubberz.”
HumanStandard initially did not classify the song as AI-generated. After adding approximately 1,000 Treblo-generated tracks to retrain its detection model, however, the service tested the song again and classified it as entirely AI-generated.
This still does not provide conclusive proof. HumanStandard’s findings have not yet been reproduced by another independent organization using the same methodology, and Treblo’s founder has not confirmed that the song originated from the platform.
Fenix Flexin has continued to deny that generative AI was involved. He has said that the differences in the recording came from Auto-Tune, reverb and his deliberate use of a British accent.
Why Do Live Performances and Production Footage Still Fail to Prove the Song’s Origin?
Fenix Flexin has publicly shared what he described as footage of the song’s Pro Tools session and has performed the track several times in live or studio settings.
Critics have pointed out, however, that some of these performances relied heavily on prerecorded vocals. They also argue that he has not consistently reproduced the vocal range, accent or lyrics heard in the released recording.
On the other hand, footage of a Pro Tools session can only demonstrate that audio was imported into or edited within the software. It cannot prove how the original vocals, melody or lyrics were created.
Unless the creators release the original recordings, unprocessed vocal tracks, complete production history or records from a generative platform, the available evidence remains insufficient to prove that the song was performed entirely by a human. It is also insufficient to prove that the entire track was directly generated by AI.
Why Is It So Difficult to Reach a Verdict Using AI Music Detectors?
AI music-detection tools generally need to have been trained on output from a particular generative model before they can identify the audio patterns associated with that model.
HumanStandard initially lacked Treblo samples and therefore failed to identify the song as AI-generated. Its result changed substantially only after the model was retrained.
This means that when a detector does not classify a song as AI-generated, the result only indicates that the tool did not find characteristics it recognized. It does not prove that the work was created entirely by humans.
The Verge submitted “Rubberz” to five AI music-detection tools. Most returned probabilities of approximately 20% to 30% for AI generation or AI assistance.
These findings were inconsistent with HumanStandard’s classification of the track as fully AI-generated. The disagreement illustrates that current detection tools do not yet share a consistent technical standard.
What Are the Weaknesses of SIQA’s Self-Reporting System?
Sonic Intelligence Academy currently operates AI music rankings including the Top 100 AI Songs chart. However, inclusion is based mainly on disclosures submitted by artists or distributors rather than technical analysis of every released song.
SIQA reportedly told Zinstrel that if “Rubberz” were submitted as an AI-generated track, its performance could place it at No. 1 or No. 2 on the chart.
Because Fenix Flexin’s team denies that AI was involved, the song will not be included.
This system can identify works whose creators are willing to disclose AI use, but it cannot detect songs whose creators refuse to report or deliberately conceal the use of AI.
Do Spotify and Apple Music Already Label AI-Generated Music?
Spotify and Apple Music have both begun developing systems for disclosing the use of AI.
In 2026, Apple Music introduced voluntary Transparency Tags covering the use of AI in recordings, songwriting, cover artwork and music videos. Spotify uses DDEX metadata standards and song credits to present information about AI involvement.
At present, however, these systems still rely largely on record labels, distributors and creators to submit the information themselves. The platforms do not conduct a comprehensive technical investigation of every uploaded song.
As a result, listeners may still see no indication of AI involvement when a work has not been voluntarily labeled.
Spotify also has a policy prohibiting unauthorized voice impersonation. That policy concerns the misuse of an artist’s identity and vocal rights. It does not provide a system for determining the complete production origin of a song.
How Should Creators and Listeners Evaluate AI Music Controversies?
Creators who use generative AI during songwriting, melody creation, vocal production, arrangement or post-production should decide what they intend to disclose before releasing the work.
They should also preserve original recordings, revision histories and production-session files. These materials can help answer public questions and provide evidence if disputes arise involving voice rights, copyright or revenue sharing.
Listeners should not draw conclusions based solely on one live performance, a single detection tool or a voice comparison circulating on social media.
Charts such as SIQA can only confirm which works have been voluntarily reported as involving AI. They cannot prove that songs absent from the chart were produced entirely without AI.
A more complete evaluation requires comparing several forms of information:
- Statements from the artist and producer
- Official production and songwriting credits
- Original recordings and production-session records
- Results from multiple detection tools
- AI-related labels displayed by streaming platforms
When these sources contradict one another, the most accurate conclusion remains “currently unconfirmed,” rather than whichever answer best supports a listener’s existing position.
At present, “Rubberz” cannot be reliably classified as a fully human-created work, an AI-assisted work or a completely AI-generated work.
Listening impressions alone are not enough to settle the question. Detection tools can produce false negatives, while self-reporting systems can be incomplete or deliberately circumvented.
The controversy ultimately exposes a broader problem: the current music industry still cannot reliably answer a seemingly simple question—who created a song, and which tools were used to make it?
Frequently Asked Questions
No. The question remains unresolved.
Fenix Flexin and producer Purps on the Beat continue to deny that AI was involved. However, a number of music critics, media outlets and listeners have argued that the song’s vocals and structure strongly resemble output produced by current AI music-generation tools.
Treblo is a free AI music-generation platform.
Users can enter the names of singers or artists to generate music that imitates aspects of their vocal characteristics or style. This functionality currently exists in a legally uncertain area, and Treblo is not an officially licensed artist-voice platform.
No. The Billboard Hot 100 does not currently require songs to disclose whether AI was used.
A song may qualify for the chart as long as it meets the applicable streaming, sales and radio-airplay requirements.
There is currently no clear precedent indicating that Billboard would remove a charting song solely because generative AI had been used.
Technical and disclosure standards for identifying AI-generated music have not yet been fully established.
A song could potentially face removal from a streaming service if it violated policies involving copyright, unauthorized voice imitation or fraud, but the use of AI alone would not necessarily result in removal.
Yes. SIQA publishes the Top 100 AI Songs chart, which is designed to track music involving AI.
However, the chart relies on self-reporting. Works whose creators do not disclose AI use will not be included.