Anthropic CEO Dario Amodei: Cancer and other diseases may become history in less than 10 years
Anthropic CEO Dario Amodei said in the company’s IPO roadshow video that cancer, Alzheimer’s disease, heart disease, and mental illness, “I believe, will become a thing of the past in less than ten years.” He also pointed out that only companies that properly manage AI risks and earn society’s trust may obtain the “operating license” to continue developing and applying the technology; if handled properly, humanity may enter a “miraculous era” of solving problems that have lasted for thousands of years.
This statement continues the framework from his article “Machines of Loving Grace,” published in October 2024. In that article, he proposed a “compressed 21st century”: after powerful AI emerges—AI with capabilities beyond the top scientists in most fields and able to autonomously complete long-term tasks—the biological and medical progress that would originally take humans 50 to 100 years could be compressed into 5 to 10 years. Specific ideas include eliminating most cancers (with incidence and mortality possibly reduced by more than 95%), preventing Alzheimer’s disease, and extending the human average lifespan from about 75 years to about 150 years. In his article “We Must Pace the Frontier” in September 2026, he again said that AI is expected to cure most major diseases in the next 5 to 10 years.
Amodei’s personal experience is related to this. His father died of hepatitis C in 2006, and several years later a new generation of antiviral drugs emerged with a cure rate of about 95%. He himself had early-stage testicular cancer and recovered after chemotherapy. In his September 2026 article, he mentioned that these experiences made him feel the significance for patients of medical breakthroughs arriving “a few years earlier.”
On October 27, 2026, Anthropic disclosed in an official research update that it had established drug development collaborations with six organizations, including Bristol Myers Squibb, and acquired Mithril Medicine, which had only one full-time employee. Founder Cody Tranbarger used Claude to coordinate experiments, data, and contracts across 11 external laboratories, and in less than 18 months advanced a potential therapy for a rare muscle-wasting disease to mouse experiments; the grip strength and muscle strength of some treated mice reached about twice those of the untreated group, and survival improved. Anthropic said that, if subsequent progress goes smoothly, the related drug could enter human clinical trials as early as the end of 2028. The company also emphasized that the work is still mostly at the preclinical stage.
Amodei has previously also pointed out that clinical trials and regulatory approval are the main bottlenecks to acceleration. New drugs typically require about 7 to 8 years to pass U.S. and European regulation. He advocates shortening the development cycle through better AI simulation, toxicity prediction, and biomarkers, while retaining necessary safety validation. AI can speed up molecular design and data analysis, but it cannot directly compress the time needed for the human body to produce efficacy or reveal long-term side effects, especially for chronic diseases such as Alzheimer’s disease.
Original source:
Title: Early progress on accelerating drug development with Claude Publisher: Anthropic Publication date: 2026-10-27 URL: https://www.anthropic.com/research/early-progress-drug-development Source type: official announcement Verification status: full text actually read for this report (as of 2026-11-11), no access restrictions Supporting content: supports Anthropic’s collaboration with six organizations, including Bristol Myers Squibb; the acquisition of Mithril Medicine; Cody Tranbarger’s coordination of 11 laboratories with Claude; the progress of a potential therapy for a rare muscle-wasting disease to mouse experiments in less than 18 months and the observation of roughly twofold improvements in strength; the earliest expected entry into human trials at the end of 2028; and the fact that the work is still mostly preclinical. It does not prove that the disease “becoming history” prediction made at the roadshow has been realized, nor does it prove that the timeline for clinical trials has already been shortened.
ABAB AI Insight
Dario Amodei’s medical predictions are directly related to his personal experience. He was diagnosed with early-stage testicular cancer in 2014 and recovered after chemotherapy; his father Riccardo died of hepatitis C in 2006, and several years later drugs such as sofosbuvir raised cure rates for most patients to about 95%. In “Machines of Loving Grace” in 2024, he explicitly made “compressing 50-100 years of biological progress into 5-10 years” the core framework, and specifically named cancer (with incidence and mortality possibly reduced by 95% or more) and Alzheimer’s prevention. In “We Must Pace the Frontier” in September 2026, he again reiterated that “most major diseases will be cured in the next 5-10 years” and used his personal experience as a source of “urgency.”
On the capital and business side, Anthropic is moving Claude from a general-purpose model toward specific drug pipelines. In October 2026, the company disclosed in its own research update collaborations with six organizations, including BigHat Biosciences, Bristol Myers Squibb, and Generate Biomedicines, and also acquired Mithril Medicine. Tranbarger, as the only full-time employee, used Claude to handle contracts, logistics, and coordination with 11 external laboratories, advancing a potential antibody therapy for a LAMA2-related congenital muscular dystrophy from concept to positive mouse results, with some indicators improved by about two times. Anthropic said that, if progress continues smoothly, human trials could begin as early as the end of 2028. This is not simply selling APIs to the outside world, but embedding the model into specific links from target evaluation and molecular design to trial planning, while still retaining the clear limitation of the preclinical stage.
Compared with historical cases, the emergence of hepatitis C cure drugs in the 2010s was a real medical acceleration: the time from discovery to high-cure-rate deployment was relatively short, but the overall pipeline was still constrained by trials and approval. Amodei himself acknowledged in his article that average cancer drugs only extend survival by months and side effects need to be measured carefully, that Alzheimer’s disease is similar, and that large trials and regulatory judgment remain bottlenecks. The similarity with the hepatitis C case here is that “when new tools greatly raise the success rate, deployment speeds up”; the key difference is that Claude currently mainly accelerates design and coordination (such as 35,000 antibody designs and 1,010 variant tests), while the timing of biological effects and validation of long-term safety cannot be directly compressed by the model.
Therefore, Amodei’s prediction changes Anthropic’s narrative of “operating license” and its specific investment direction: it binds AI safety to “earning society’s trust in order to continue development,” while bringing Claude into real drug programs through the Mithril acquisition and six collaborations. Observable indicators are how many of these preclinical projects can enter human trials according to the company’s expected timeline, and whether there is measurable time compression in trial design and regulatory document preparation (for example, in the Enveda case, document review falling from more than 25 hours to less than 3 hours). In the absence of completed human efficacy data, this path still depends on whether subsequent trials and approvals proceed as expected.
ABAB News · Cognitive Law
Models can compress design and coordination time, but cannot replace the time needed for human verification of efficacy and side effects.
Personal experience can shape a sense of urgency about “a few years earlier,” but predictions must still retain clinical and regulatory conditions.
Binding safety to “operating license” makes the medical vision simultaneously a condition for the company’s continued operation.