Those AI-built viruses are bacteriophages, viruses that infect bacteria rather than humans, making them useful test cases but also vivid warnings.<\/p>\n Why experts fear AI viruses<\/p>\n The work revealing how AI-designed proteins can slip past genetic safety checks was led by Bruce J. Wittmann at Microsoft Research<\/a>.\u00a0<\/p>\n He works as a senior applied scientist, focusing on tools that make DNA screening more reliable for laboratories and synthesis companies.<\/p>\n These projects rely on genome-language models, AI<\/a> tools that guess plausible stretches of DNA in a sequence.<\/p>\n Once trained on thousands of sequences, those models can suggest entirely new genomes that still resemble natural viral families.<\/p>\n That creative reach makes it harder for experts to anticipate how a machine-designed virus or protein might behave in real experiments.<\/p>\n In a recent preprint, researchers used those models to design hundreds of candidate phage genomes and successfully grew 16 working viruses.<\/p>\n Because each phage infects only specific bacteria, doctors hope tailored cocktails can treat stubborn infections while sparing helpful microbes and human cells.<\/p>\n Clinical reviews<\/a> describe patients with antibiotic-resistant infections who improved after receiving experimental phage therapy when standard drugs had failed.<\/p>\n Dual-use dilemma<\/p>\n Scientists use the term dual-use research, work that can help or harm depending on intent, to describe such powerful technologies.<\/p>\n The same algorithm that optimizes drug droplets for asthma inhalers could also reveal recipes for more efficient aerosolized chemical or biological attacks.<\/p>\n The models are considered capable, particularly when tested on real-world health data rather than controlled examples.<\/p>\n Researchers argue that carefully limiting which genomes<\/a> and toxins appear in training data is an essential first safety barrier.<\/p>\n How AI viruses evade screenings<\/p>\n Many DNA synthesis firms screen nucleic acids, genetic molecules that carry hereditary information, and flag customer orders that match pathogens or toxins.<\/p>\n Wittmann\u2019s team showed that advanced protein design tools can rewrite dangerous proteins into new sequences that still work yet evade standard screening filters.<\/p>\n Working with several DNA<\/a> suppliers, they developed new algorithms that focus on protein structure and function, sharply improving detection of these disguised sequences.<\/p>\n Building better defenses<\/p>\n Those patches are now being integrated into commercial screening pipelines, transforming a hidden weakness into a playbook for defending against AI-assisted design.<\/p>\n Effective DNA synthesis screening, automated checks on gene orders and customers, gives authorities a chance to stop dangerous projects before material is produced.<\/p>\n The International Biosecurity and Biosafety Initiative for Science promotes<\/a> screening standards so companies and regulators judge genetic orders with similar criteria.<\/p>\n Federal policies for safety<\/p>\n A federal framework<\/a> links research funding to nucleic-acid screening, encouraging agencies to favor labs that buy DNA from vetted providers.<\/p>\n Under this policy, providers are expected to screen every order, assess customers, report suspicious requests, and keep records for years.<\/p>\n Future administrations may revise or replace these expectations, but they demonstrate how governments can turn safety guidance into concrete purchasing conditions.<\/p>\n Similar requirements could eventually apply to high capability AI developers, tying access to public contracts to documented testing and clear internal safeguards.<\/p>\n Global efforts for safety standards<\/p>\n The International Biosecurity and Biosafety Initiative for Science brings together companies and regulators to update screening norms as AI and synthesis technologies change.<\/p>\n One important partner is the International Gene Synthesis Consortium, whose members commit to screening sequences and customer identities using a harmonized protocol.<\/p>\n The United Kingdom has created an AI-Safety institute<\/a> to test models, evaluate risks, and share methods for reducing misuse.<\/p>\n Tessa Alexanian, a researcher involved in the study, described the approach as deliberately flexible. <\/p>\n \u201cThis managed-access program is an experiment and we\u2019re very eager to evolve our approach,\u201d she said.<\/p>\n Practical limits to bioweapons<\/p>\n For all the worry, there remains a wide gap between digital genome design and reliably engineering contagious viruses that can spread among humans.<\/p>\n The recent phage work targeted tiny bacterial viruses, far simpler than the complex pathogens responsible for diseases such as influenza or COVID-19.<\/p>\n Even with good designs, scientists still need high-containment facilities, careful oversight, and months or years of experimentation to produce stable, predictable organisms.<\/p>\n Experts worry because advances in automation, DNA<\/a> synthesis, and modeling are shrinking these obstacles, lowering the effort required to attempt dangerous projects.<\/p>\n AI viruses and the future<\/p>\n More researchers now argue for biosecurity, policies that prevent misuse of biological tools, combining better training choices with simple screening of outputs.<\/p>\n Some proposals look further downstream, suggesting environmental surveillance that monitors sewage, air filters, or hospital samples for genetic traces of unauthorized production.<\/p>\n At the same time, these AI<\/a> tools could accelerate discovery of antibiotics, vaccines, and tailored phage therapies that rescue patients when drugs fail.<\/p>\n Many experts see a shared responsibility for funders, journals, companies, and universities to require safety evaluations whenever powerful AI tools touch biological work.<\/p>\n The study is published in the journal Science<\/a>.<\/p>\n \u2014\u2013<\/p>\n Like what you read? Subscribe to our newsletter<\/a> for engaging articles, exclusive content, and the latest updates.<\/p>\n Check us out on EarthSnap<\/a>, a free app brought to you by Eric Ralls<\/a> and Earth.com.<\/p>\n \u2014\u2013<\/p>\n","protected":false},"excerpt":{"rendered":"Scientists have now used artificial intelligence, computer systems that learn patterns from data, to write complete viral genomes…\n","protected":false},"author":2,"featured_media":394555,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[256,254,255,64,63,128,105],"class_list":{"0":"post-394554","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-artificial-intelligence","8":"tag-ai","9":"tag-artificial-intelligence","10":"tag-artificialintelligence","11":"tag-au","12":"tag-australia","13":"tag-science","14":"tag-technology"},"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts\/394554","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/comments?post=394554"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts\/394554\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/media\/394555"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/media?parent=394554"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/categories?post=394554"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/tags?post=394554"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
\n ![\"EarthSnap\"\/](\"https:\/\/www.newsbeep.com\/au\/wp-content\/uploads\/2025\/12\/1766650691_946_earthsnap-banner-news.webp.webp\")
\n<\/a><\/p>\n