There are real world demonstrations of how autonomous artificial intelligence agents are poised to disrupt biomedical research, according to two papers published May 19 in Nature. Each describes an AI system that assists across the piece, from generating hypotheses to designing experiments, analyzing the data and refining hypotheses in the light of new data.

On Sunday, May 17th, 2026, the World Health Organization classified the ongoing Bundibugyo ebolavirus outbreak in the Democratic Republic of Congo (DRC) as a public health emergency of international concern (PHEIC). The rapid escalation to PHEIC is due to several factors. Given the high number of cases, the outbreak has likely been going undetected for some time, and may be a “much larger outbreak than what is currently being detected and reported, with significant local and regional risk of spread,” according to the WHO statement. The outbreak appears to already have crossed the border from the DRC into Uganda at least twice. And all this is happening with a virus for which there are no approved treatments or vaccines.

Directed evolution has become a central pillar in gene therapy. This engineering strategy enables the generation of more efficient variants of genetic editors and delivery vectors. Molecular diversification methods are increasingly sophisticated and are now accelerated by machine learning and AI tools, as showcased at the 29th Annual Meeting of the American Society of Gene and Cell Therapy (ASGCT) held in Boston this week.

Gene therapies rely on vectors to reach the target tissue where they act, such as adeno-associated viruses (AAVs) or lipid nanoparticles (LNPs), among other delivery strategies. Each combination is optimized for a specific cell type and indication, aiming to overcome challenges such as efficacy, specificity and toxicity. On May 13, 2026, two sessions included in the scientific symposia of the 29th Annual Meeting of the American Society of Gene and Cell Therapy (ASGCT), being held in Boston this week, addressed AAV-related toxicities, which have led to fatal cases in clinical trials and remain an area for improvement in approved therapies.

Circular RNA (circRNA) is not a new concept, but it is a novel strategy in the field of gene and cell therapy. While mRNA vaccines have revolutionized medicine, this RNA fragment without free ends surpasses their performance in both efficacy and durability, bringing it to the attention of several pioneering companies. The latest advances in circRNA presented at the 29th Annual Meeting of the American Society of Gene and Cell Therapy (ASGCT) clearly surpass the performance achieved with linear mRNA.

A designed chimeric virus induced broadly neutralizing antibodies against the macaque equivalent of HIV. The strategy works in two steps: first it uses an envelope protein with a mutation that reduces the glycan shield that makes it invisible to the immune system, and then it exposes the part of the protein most likely to generate these antibodies capable of blocking many variants of the virus. The macaques developed potent and diverse antibodies with this approach, which pave the way for the development of an HIV-1 vaccine.

The initial appraisal of the first complete genome sequence of a hantavirus isolated from a patient in Switzerland who was a passenger on the cruise ship MV Hondius is consistent with a spillover from its natural reservoir, rather than the emergence of a markedly altered virus.

Researchers at Daping Hospital in China have reported that liver-targeted delivery of the APOE3-Christchurch (APOE3Ch) variant, a rare protective form of apolipoprotein E, can indirectly reduce brain pathology, highlighting the therapeutic potential of peripheral approaches to Alzheimer’s disease.

A new mRNA and lipid nanoparticle (mRNA-LNP) platform could selectively reprogram in vivo cytotoxic effector T cells (Teff), the cells responsible for eliminating infected or tumor cells. To achieve this, scientists at the University of Pennsylvania conjugated LNPs with fractalkine, a molecule that binds to the CX3CR1 receptor, which is a marker of Teff cells. Using this strategy, the researchers delivered an mRNA encoding new proteins such as IL‑2 or human CD62 L‑selectin, opening the door to temporarily reprogramming these cells within the body, both in the blood and in lymphoid tissue, where they reside and become activated.

Scientists at the La Jolla Institute for Immunology have identified and characterized human antibodies that neutralize the measles virus by blocking its entry into the cell. This is the first time that antibodies have been shown to bind effectively to two essential viral proteins, creating a dual blockade that prevents infection. Unlike the current vaccine, which is based on an attenuated virus and is not recommended for immunocompromised individuals, these monoclonal antibodies could be used both as a new vaccine approach and as a treatment for the entire population.