[Digital Today reporter Choi Jae-won] U.S. AI company Anthropic has acquired a biotech startup that was run by a single employee with the help of AI and has entered rare disease drug development. It aims to change pharmaceutical research and development by using AI from drug candidate discovery to experiment management.

On Oct. 28 (local time), Japanese outlet ITmedia reported that Anthropic said it has acquired Mithril Medicine, which develops treatments for pediatric rare muscular dystrophy, and is pursuing in-house drug development.

Mithril Medicine, acquired in July, was a company where founder Cody Tranbarger (코디 트런버거) was the only full-time employee. Tranbarger used Anthropic's AI model Claude to manage contracts with 11 external research institutions, experiment schedules, logistics and data analysis.

Using this approach, Mithril Medicine advanced treatment ideas to the animal-testing stage in less than 18 months. The research targets are LAMA2-related congenital muscular dystrophy and alpha-dystroglycanopathy, both rare diseases that cause severe impairment of muscle function.

The research team developed a bispecific antibody that binds to two types of proteins at the same time to restore the function of linking muscle fibers and surrounding tissue.

After administering a prototype antibody for 6 weeks to mice with a LAMA2-related disease, some test groups had grip strength and muscle strength about 2 times higher than the control group. Survival rates also improved, but did not recover to the level of normal mice.

Claude was also used in follow-up research. The research team had the AI analyze 341 pages of manufacturing data on 113 candidate antibodies, and Claude classified test targets and candidates requiring additional purification in less than 8 minutes.

Errors were also confirmed. In emails written by Claude, there were cases where antibody quantities were recorded incorrectly, and at least 3 cases were sent to external research institutions even after human review.

Anthropic views this research as the first example of a "one-person drug development pipeline" model. Tranbarger has continued the research after joining and is preparing to develop new treatments for rare diseases.

The company expected that if verification of candidate materials proceeds smoothly, it could begin human clinical trials as early as the second half of 2028. It said the current results are early animal-test findings, and safety and therapeutic effects in humans have not been verified.