Suno Visuals turns songs into video, priced in separate credits

Two consumer AI products moved today: one gives your song a face, and one takes your screen back. Both are really about who gets to read what.
Suno's new Visuals tool turns a song into a clip of 5 to 15 seconds, and bills it in a second currency you have to buy separately. The music platform launched Visuals on Tuesday across web and mobile, adding video and image generation to the same place you write songs, according to Suno's announcement. Clips lip-sync to the vocal, can be extended and reordered on a timeline, and come with lyric videos, album art and a choice of aspect ratio. You can pick one of Suno's original characters or upload a reference image, then prompt against the song's structure so the visuals land on the chorus instead of drifting over it.
The interesting part is the model underneath. Suno calls it H3 Move, and its own FAQ is blunt about where it came from: "The model is called H3 Move, based on H3 from MiniMax, which we optimized for our community including adding additional protections and safeguards." That makes a US consumer product — one with more than 100 million users and a marketing budget — a downstream customer of a Chinese video model that ships with an unusually narrow open-weights licence.
MiniMax released H3 weights on August 3, and the licence excludes the United States, the European Union, the United Kingdom and South Korea from self-hosting, as Tech Times reported at release. Suno is a US company. So it is either calling MiniMax's hosted API or holding a separate commercial licence — and Suno has not said which. That silence is the sort of detail that reads as housekeeping until a plaintiff's lawyer asks the same question under oath.
Pricing is the part most people will search for, and the answer is that video does not come out of your song credits. Visual credits are a separate purchase; packs do not expire, and the cost of a clip varies with its length and whether you render in Fast or Standard mode. The tiers matter more than the pack prices, because the free plan gets roughly one five-second video before it hits the cap. Pro at $10 a month and Premier at $30 a month get a temporary allocation of visual credits to try the feature; Platinum, at $120 a month or $100 a month billed annually, bundles 20,000 visual credits a month, per Suno's pricing page. For anyone doing this weekly, the video budget now dwarfs the music budget.
Suno's chief product officer, Jack Brody, frames the launch as a cost-and-speed story: video generation was previously too slow and too expensive to hand to ordinary users, and the models have only now crossed the line. What he does not pretend is that the tool is neutral. "Prompts that work on other video platforms may not work on Suno and that's by design," he wrote. In practice that means text and visual filters aimed at named artists and public figures, plus a watermark embedded before anything is shared off-platform.
That design is not caution for its own sake. MiniMax is being sued by Disney, Universal and Warner Bros. Discovery, and a judge refused to dismiss that case in May. Suno itself is being sued by Universal Music Group and Sony Music over its label-backed v6 model — we covered that filing in September, Sony and UMG sue Suno over its label-backed v6 model. Putting a video generator on top of a model from a company fighting Hollywood studios, inside a product already fighting record labels, is a choice that stacks two litigation queues in one workflow.
The generous reading is that the licensing risk sits upstream: if MiniMax's weights are the problem, that is MiniMax's problem, and Suno's terms push responsibility for generated output onto the user, as they do for every generator. The less generous reading is that Suno just made its own product more legally interesting at the exact moment it is trying to convince labels it can be a licensed partner rather than a defendant. Note which way the technical dependency runs: Suno did not build the video model, it adapted one. The base layer, and the exposure that comes with it, belongs to someone else.
Signal has started blocking screenshots inside its app so that Siri, Gemini and other operating-system assistants cannot read your messages off the screen. The change, announced in a blog post by Signal president Meredith Whittaker on Friday, applies on iOS 27 and on Android versions where the app can detect an assistant trying to read Signal's screen, Signal said. It covers the app switcher previews as well, and it is on by default — users can turn it back on in settings (iOS labels it "Block Siri AI", Android calls it "Screen security"), but the warning is explicit that doing so may expose your contacts' messages too.
Signal is unusually direct about why. Whittaker argues that operating-system AI agents are "undermining Signal's ability to independently guarantee privacy," and that the broader problem is structural: "it is a serious problem that Signal — and application developers beyond — lack robust tools to prevent AI agents from undermining application-level privacy." She calls Signal "the canary in the coal mine" and asks Apple, Google and Microsoft for simple technical controls and documentation so any app can draw the line, not just the one with a security team large enough to reverse-engineer the OS.
The fix is honest about its own limits. Signal says the protection is incomplete, because agents can still reach app data through the device filesystem, and because deep accessibility permissions let an agent read text directly without ever needing a pixel — the company names Meta's Muse as an example of permissions used like a skeleton key. On macOS, Signal cannot implement the block at all and points users to the Excluded Apps setting instead. That setting exists because Apple shipped it in iOS 27.2 and macOS 27 in late October, per EFF — a general control that switches off Siri's onscreen awareness for a chosen app. Signal's screenshot toggle is the same idea, applied one app at a time, because the platform would not do it for everyone.
What to watch: whether Apple or Google ships a developer-facing API for agent data access before a second messenger follows Signal. One app blocking screenshots is a product decision. Three would be a platform problem.
Would you give up screenshots in your messaging app to be sure no agent reads your chat? Write to us via our contact page.







