Researchers at UCSF reported a method to generate CAR T cells directly inside living hosts by delivering a large DNA sequence and achieving site-specific integration with a dual‑vector approach. Justin Eyquem and collaborators demonstrated stable, locus‑specific insertion that produced functional CAR T cells in humanized mice and cleared aggressive blood and solid tumors. The lead sentence: the team claims the first programmable, site‑specific integration of a large DNA payload into T cells in vivo. The study showed the method outperformed random viral integration approaches and produced durable antitumor responses in preclinical models. Tests included humanized immune system mice and multiple tumor types; authors highlighted superior trafficking and expansion of edited cells versus prior transient or random‑integration in vivo strategies. "In vivo CAR‑T" refers to engineering T cells inside the patient rather than manufacturing them ex vivo, a step that could eliminate weeks‑long centralized manufacturing. Investigators reported therapeutic activity in leukemias, myeloma and surprising efficacy in solid tumor models — a major technical hurdle for cell therapies. The technology remains preclinical; safety, off‑target editing and scalability must be addressed before human trials. The paper positions programmable, site‑specific integration as a potential path to simplify access and cut costs of engineered cell therapies.