OpenAI’s Sora is shutting down. Splice just launched AI tools that actually pay the humans whose work they build on. ElevenLabs dropped a free music generation app. And galleries still don’t think AI art is real art.
The creative AI space is splitting into two very different camps: companies that treat creators as raw material, and companies trying to build something creators might actually want to use.
The Sora Collapse
The biggest story in creative AI right now isn’t a launch — it’s a shutdown.
OpenAI confirmed that Sora, its much-hyped text-to-video generator, is being discontinued. The Sora app closes April 26, with the API following in September. The numbers behind the decision are staggering: Sora was burning an estimated $15 million per day in compute costs while generating just $2.1 million in total lifetime revenue. Downloads had dropped 66% from their November 2025 peak.
Then there was the Disney debacle. Disney had committed $1 billion to a Sora partnership, only to learn about the shutdown less than an hour before the public announcement. No money ever changed hands.
The takeaway for creators who were building workflows around Sora: platform risk is real, even when the platform is backed by the most well-funded AI company on Earth.
Who’s Filling the Void
Sora’s exit leaves a crowded field of competitors fighting for video creators’ attention.
Kling AI has emerged as the frontrunner. Kuaishou shipped three major updates between January and March, culminating in Kling 3.0 with 4K output and six-axis camera control. It generates up to two minutes of video — nearly five times Sora’s 25-second cap — making it viable for product demos, training content, and extended social clips.
Google Veo 3.1 launched in February with native synchronized audio, a feature that went from research paper to production in under 12 months. Four of the six major AI video models now generate audio natively, up from zero in early 2025.
Runway remains the choice for professional production. Gen-4 Turbo delivers the best character persistence and temporal consistency in the market, though it hasn’t shipped a major update since Gen-4.5 in November.
The practical strategy for agencies and creators in 2026: don’t commit to one platform. Use Runway for hero content, Kling for volume, and Pika for social experimentation.
Splice Gets Compensation Right
While most AI companies treat creative works as training data and move on, Splice is trying something different.
On April 15, Splice announced three new generative AI tools — Variations, Craft, and Magic Fit — designed to modify and customize the three million-plus samples in its library. The tools don’t generate from scratch. They transform existing human-made samples.
Here’s what matters: creators get paid. Every time someone uses a sample as a source, and every time an AI-generated variation gets downloaded, the original creator receives compensation. Variations remain fully traceable to their source.
Variations generates five different versions of a chosen sound while preserving the original’s identity. You specify key, tempo, and complexity. Craft transforms samples into playable instruments inside Splice’s plugin. Magic Fit, coming this summer, will adapt sounds to match the harmonic and rhythmic context of your session.
This is worth watching because it’s a model that could work: AI as augmentation of human creativity, with a compensation trail that actually functions.
ElevenLabs Enters the Music War
ElevenLabs, fresh off a $500 million Series C at an $11 billion valuation, launched ElevenMusic on April 1. The iOS app lets you generate up to seven songs per day for free using text prompts. A $9.99/month Pro tier unlocks 500 tracks monthly.
What sets ElevenMusic apart from Suno and Udio is its licensing foundation. The platform was built on exclusive partnerships with the Merlin Network and Kobalt Music Group — making it the only major music generation platform offering commercial-use tracks from day one.
Suno, meanwhile, continues to dominate on sheer scale with nearly 100 million users and a $2.45 billion valuation. But its licensing remains murkier, and it faces ongoing lawsuits from major labels.
For independent musicians, the question isn’t whether to use these tools. A Symphonic survey found 87% of music creators have already integrated AI into at least one part of their workflow — from generating ideas and mastering tracks to creating marketing materials. The tools getting the most real-world use aren’t the flashy generators. They’re stem separators like Moises.ai, mastering tools, and content schedulers.
The Art World Isn’t Convinced
If you’re hoping the gallery world has warmed up to AI art, the data says otherwise.
Artsy’s inaugural AI Survey, polling more than 300 gallery professionals, found deep skepticism. 41% of galleries say AI “rarely comes up” with collectors. 16% report collectors actively avoid AI-assisted artworks. Only 15% have seen “curiosity-driven interest” — questions without purchases.
The survey found that 28% of respondents can’t even agree on what counts as “AI art.” And the vast majority of gallery-represented artists aren’t using AI in their practice. Many are openly critical of the ethical implications.
Where galleries are adopting AI is behind the scenes: streamlining communications, research, planning, and admin tasks. The Artsy survey’s blunt conclusion: AI’s near-term impact on the commercial art world will be operational, not artistic.
Writers: Cautious but Pragmatic
Fiction authors have reached their own equilibrium. The Authors Guild found 82% of writers worry about AI homogenizing literary style — but those fears apply mainly to writers who use AI passively, accepting unedited output.
The most productive fiction writers in 2026 typically use two tools: one chatbot for creative brainstorming and one specialized platform (Sudowrite, NovelCrafter, or NovelAI) for long-form execution. 73% say AI helps them overcome writer’s block. None of them are asking AI to write their books.
What This Means
The creative AI market is sorting itself out along a clear axis: tools that replace creators versus tools that serve them.
Sora collapsed because it was a tech demo dressed up as a product — expensive to run, legally precarious, and disconnected from how professionals actually work. Splice is investing in traceability and compensation. ElevenLabs built licensing partnerships before launching. These aren’t just nicer approaches; they’re more sustainable business models.
The gallery world’s resistance points to something real: creative communities don’t adopt tools because they’re impressive. They adopt tools that respect the work that came before.
What You Can Do
If you were using Sora: Export your work before April 26. Evaluate Kling for longer-form content, Runway for quality-critical work, or Veo for projects that need synchronized audio.
If you make music: Look at Splice’s new tools if you’re already in their ecosystem — the compensation model sets a standard worth supporting. ElevenMusic is worth testing for quick demos, but check the licensing terms carefully before commercial use.
If you’re a visual artist: The Artsy survey confirms what most artists already know — AI hasn’t changed the gallery market yet. Focus on your practice. If you want to experiment with AI as a tool, keep it in your sketchbook phase, not your finished work.
If you write fiction: Use AI for brainstorming and structure, not prose. The tools are genuinely helpful for breaking through blocks, but the 82% of authors worried about homogenization aren’t wrong.