ByteDance unveils AI-designed drug candidate targeting IL-17 for autoimmune diseases

ByteDance has publicly shown for the first time how its AI is moving beyond video recommendations and into biomedicine. Its Anew Labs unit has presented a drug candidate for autoimmune diseases, aimed at a target that pharma has spent decades treating as too difficult for conventional small molecules.

What was shown in Boston

On April 16, 2026, at the annual meeting of the American Association of Immunologists in Boston, the Anew Labs team disclosed preclinical data on its lead candidate: an oral small molecule created with generative AI to block IL-17. This signaling protein is linked to psoriasis, rheumatoid arthritis and ankylosing spondylitis. For ByteDance, it is the first public sign that its drug discovery effort has moved beyond internal research and now has a concrete therapeutic project.

What matters most is not the company's name but the type of target. In this form, IL-17 belongs to protein-protein interactions that have long been called undruggable: too "flat" and large for classical small molecules. According to the talk's abstract, the Anew candidate can act on several IL-17A and IL-17F dimers at once, and in preclinical models showed activity comparable to the antibody bimekizumab. It is not a drug on the market but an early candidate, yet steps like this are considered the real test for AI approaches in pharma.

"These are the first oral small molecules with strong action on IL-17F

and pronounced in vivo activity," the abstract says.

Why it matters

Today the most successful anti-IL-17 drugs are mostly injectable biologics. They work, but they require injections, are harder to manufacture and are not always convenient for patients. If ByteDance can actually take an oral molecule into the clinic, and then to market, it would be more than another AI case study: it would be an attempt to replace part of the expensive injectable regimens with a pill. For autoimmune diseases this is a particularly sensitive scenario, because therapy is often needed for a long time.

• Target: IL-17A and IL-17F, key drivers of several autoimmune diseases
• Candidate format: an oral small molecule, not an injectable antibody
• Design was carried out through AI-enabled structure-based generative design and virtual screening
• Preclinical data were shown on binding, cellular activity and in vivo models

There are still plenty of reasons for caution. The results presented are conference and preclinical data, not clinical proof of efficacy in humans. In biopharma, years of optimization, toxicology, regulatory stages and a high probability of failure lie between a strong presentation and a real drug. So the news matters not as a finished breakthrough but as a signal: a major technology company is already showing not general promises but a specific molecule for specific biology.

Why ByteDance is doing this

Anew Labs looks less like a random side project inside a big tech company and more like a distinct ByteDance bet on computational biology. The company's website lists teams in Shanghai, Singapore and San Jose, as well as 36 key staff members and a scientific advisory board with alumni of Innovent Biologics, Amgen and Takeda. The ambition is plain: take targets where classical drug discovery stalls, and find forms of therapy that can be made more accessible and more convenient for patients.

The story also matters because ByteDance is building not a single project but a platform. In March 2026, Anew Labs published a preprint of AnewOmni, a generative system trained on more than five million biomolecular complexes. The authors claim the model can design functional molecules at different scales, from small compounds to peptides and nanobodies. Anew's website also lists AnewSampling and other tools, meaning the company is building a full stack for finding, evaluating and optimizing candidates.

Against this backdrop, ByteDance is entering an already established race in AI drug discovery alongside DeepMind, Anthropic and Insilico Medicine. The difference is that the conversation is no longer about polished demos or only about publications, but about who will be first to take a model-generated molecule through the entire development cycle. For tech companies it is a long and expensive game, but it is exactly what separates laboratory hype from a real pharma business.

What it means

ByteDance is showing that AI's next big battle is being fought not only over search, code or content, but also over physical products with a measurable biological effect. If Anew Labs can confirm its results in further studies, the market will get a strong argument that generative models are starting to create not just hypotheses but drug candidates for problems once considered nearly out of reach.