The ChatDSP controversy comes down to one question: is a $10 tool that turns typed descriptions into working Ableton devices a creative shortcut, or one more product built on technology many artists object to? Supporters say it lets producers build tools they could never have patched themselves. Critics point to the environmental cost of large AI models, the use of copyrighted material in training, and the risk of a flood of low-effort devices.
What gave the argument its heat was the person behind it. Dillon Bastan is a respected independent Max for Live developer, and after the backlash he said he’d consider pulling the device. It was still on sale in coverage from June 2026. Here’s what happened, what each side is really arguing, and how to work out where you stand.
What ChatDSP Actually Does
ChatDSP is a Max for Live device for Ableton Live. You type what you want, such as “a super lush reverb,” and an AI model builds a playable device inside your session. You can keep refining it with follow-up prompts, then save the result as a preset.
It comes as three separate devices: one for instruments, one for audio effects, and one for MIDI effects. Each generated device can have up to 48 parameters. Reviewers also noted that the code can be exported for use in standalone Max, or as C++ for other projects.
The $10 price isn’t the whole cost. At launch, you needed an OpenAI or Anthropic account and paid that company for usage through an API key. Bastan doesn’t sell those credits or take a cut. He also labels the tool experimental, which matters for everything that follows.
What the results look like in practice
WaveInformer ran four tests in March 2026 and published the costs, which gives a more useful picture than any marketing claim:
| Test | Model level | Reported cost | Result |
|---|---|---|---|
| Granular synth | High | $0.50 | The most successful of the four |
| 16-tap sequenced delay | High | $0.36 | Missed many requests, and some controls didn’t behave as expected |
| Microtonal MIDI arpeggiator | High | $0.93 | Failed at first and still fell short after one revision |
| Simple MIDI note repeater | Low | $0.15 | Built up through small revisions at low cost |
Those numbers are a snapshot, since model pricing changes. The pattern is more useful than the prices: small, focused requests tended to go better than long wish lists.
How the Backlash Unfolded
ChatDSP appeared in music tech news around March 10, 2026. Within about a week, MusicRadar was reporting a heated debate among Bastan’s own followers.
The reaction was sharp partly because of his reputation. He’s known for inventive, hand-built devices like Songbird, Coalescence and Natural Selection. For many followers, seeing that name on a generative AI product felt personal.
Bastan responded on Instagram and said, “I am open to taking the device down.” He called the project a fun idea rather than a serious move into AI, and asked his audience what they preferred. That’s an unusual response in a software industry that tends to defend its launches.
By June 2026, the device hadn’t gone away, but it had changed. Bastan told MusicRadar he had added support for local AI models, so users were no longer tied to large cloud providers. He also made clear that he isn’t an AI enthusiast. He described ChatDSP as one passing idea that had become technically possible to build.
What Critics Object To
The complaints aren’t one complaint. Separating them helps, because they don’t all carry the same weight.
Energy and water use
This is the most common objection, and it isn’t really about one Max device. It’s about supporting the data centers behind large AI models. The US Department of Energy reported that data centers used about 4.4% of US electricity in 2023 and projected 6.7% to 12% by 2028. That figure covers all data centers, not AI alone.
One commenter on the ChatDSP walkthrough video joked that the tool could now produce a basic Operator patch after hours of trial and error and a city’s worth of energy. It’s a fair jab at the worst case. We couldn’t find a published energy figure for generating a single device, though, so claims in either direction deserve some skepticism.
Training data and consent
AI companies have been widely criticized for training models on copyrighted work. That concern reaches beyond music, and it applies here because ChatDSP relies on those companies’ models. Some producers see any paid use of those models as backing that practice, even when the output is a reverb and not a song.
A flood of disposable devices
This may be the most practical worry. Gearnews and Synth Anatomy both raised the prospect of a wave of quickly generated commercial devices with little design thinking or long-term support behind them.
Max for Live has long had a culture of people learning from each other’s patches. If storefronts fill with untested devices, buyers lose trust and careful developers have a harder time standing out.
Lost skills and livelihoods
Building a good device takes years of learning about signal flow, synthesis and interface design. Some critics worry that skipping that work means fewer people ever learn it. Developers have a more direct concern too, since their income could shrink. Bastan himself has said he expects developer industries to change quickly.
What Supporters Argue
The case for ChatDSP starts with access. Max is a visual programming environment, and even simple ideas can take real study to build. A producer who wants one odd delay for one track usually can’t justify weeks of learning, and hiring a developer isn’t realistic for a bedroom setup.
Supporters also see it as an idea tool. You can test a concept in minutes, keep what works, and throw the rest away. For personal experiments, nobody is harmed by a rough device that only you will ever use.
Then there’s the historical argument. Bastan told MusicRadar his one real disagreement with critics is over “technical gatekeeping” and elitism, comparing some of the criticism to classical musicians rejecting keyboards, drum machines and DAWs. It’s a persuasive comparison, but it only goes so far. A drum machine never raised questions about data center demand or training data, so the concerns aren’t identical.
Interestingly, Bastan doesn’t claim the tool matches skilled human work. In his YouTube comments, he said it doesn’t yet do as well as developers who carefully craft their devices. So the strongest defense of ChatDSP is a modest one: it’s a sketchpad, not a replacement.
Where Both Sides Talk Past Each Other
Much of the argument mixes up two different questions. One is whether ChatDSP is a useful tool. The other is whether the AI industry behind it is acceptable. You can answer yes to the first and no to the second, and many people do.
| Question | Strongest “yes” | Strongest “no” |
|---|---|---|
| Is it useful? | Fast prototyping for non-programmers | Results vary, and complex requests often fail |
| Is it ethical to use? | Personal, low-volume use, with a local option | It relies on models and companies some artists oppose |
| Does it threaten developers? | Careful, supported devices still stand out | Cheap generated products could crowd them out |
| Will it hurt the community? | More people get to build and learn | Less shared craft and more forgettable devices |
The local model option is a good example of why the details matter. It may ease concerns about depending on big cloud providers, but it doesn’t settle how a model was trained. Quality expectations will also vary, so test any local setup yourself.
Bastan also offered a view that goes beyond the tool itself. He told MusicRadar that people draw different lines to reduce harm, and that it’s understandable if many draw theirs at AI. That’s a fair summary of where this debate ends up: it’s less about a Max device and more about personal limits.
Who Owns a Device Made With ChatDSP?
Few articles on this controversy cover ownership, yet it matters if you plan to sell what you generate. In the US, the Copyright Office concluded in January 2025 that prompts alone don’t provide enough control to make someone the author of AI output. Revising a prompt many times doesn’t change that on its own.
Human contributions can still count. The report says creative modifications and arrangements may be protected, decided case by case. It wasn’t written with Max devices in mind, so how it applies to a generated patch is an open question.
If you plan to sell a generated device, these are sensible steps:
- Read the terms of ChatDSP and of whichever AI provider you use.
- Keep records of what you changed by hand.
- Tell buyers how the device was made.
- Talk to an intellectual property attorney before investing real money.
This is general information and not legal advice.
How to Decide Where You Stand
You don’t need a grand position on AI to make a sensible choice. These questions will get you most of the way:
- What are you making? A throwaway effect for your own track is very different from a commercial product.
- Could an existing device do the job? The Max for Live community already offers plenty of free and paid options, and buying one supports a developer directly.
- Which concern matters most to you? If it’s cloud energy use, look at the local model route. If it’s training data, a local model may not resolve it.
- Will people rely on it? Don’t trust a generated device on stage until you’ve tested it thoroughly. Bastan’s own experimental label is a good reason for caution.
If you do try ChatDSP, start small. Ask for one clear function, test it, and then add features one at a time. That approach matches what reviewers found, and it usually costs less too.
When you buy devices, a few checks can protect you from low-effort products:
- Look for a named developer with a track record.
- Check for a real demo video showing the device in use.
- See whether the seller offers updates and support.
- Ask how it was made if the listing doesn’t say.
The Bottom Line
The ChatDSP debate has lasted because both sides have a point. It really can help non-programmers build things they couldn’t before, and the concerns about energy use, training data and device quality are real too. The most honest way to see it is as a cheap, experimental sketchpad attached to a much bigger argument about AI. Decide based on what you’re making and which concerns matter most to you, and if you sell or perform with a generated device, test it properly and be open about how it was made.
Frequently Asked Questions
Did Dillon Bastan take ChatDSP down?
Not as far as public reporting shows. He said in March 2026 that he was open to removing it, but MusicRadar still listed it for sale at $10 in June 2026. Check his store for current availability.
Why are people upset about ChatDSP?
The main objections are the environmental impact of data centers, the use of copyrighted material to train AI models, the risk of low-quality devices flooding the market, and the effect on developers’ skills and income.
Does ChatDSP need an internet connection and a paid AI account?
At launch, it needed an OpenAI or Anthropic account with paid credits. Local model support has since been added, which reduces that dependence. Check the current manual for exact setup requirements.
Are devices made with ChatDSP good enough for professional use?
Sometimes, but expect to test and refine them. Early reviews showed uneven results, with simple devices generally doing better than complex ones. Be especially careful before using one in a live set.
