+34
Leading AI companies including Anthropic, OpenAI and Google DeepMind are stepping up biological risk testing and access controls amid growing fears that advanced models could help users design viruses or bioweapons, with experts warning of a potential biological moment similar to recent AI cybersecurity disruptions.
Executives and biosecurity experts are increasingly concerned that future artificial intelligence models could help users create novel viruses or bioweapons, prompting leading AI groups to accelerate their work on biological risk testing, access controls and other safeguards.
Anthropic, OpenAI and Google DeepMind are stepping up their efforts to limit misuse of advanced AI systems, even as they hope those same systems will speed up drug discovery, aid disease research and help prevent future pandemics. The safety push reflects a growing belief among AI executives and biosecurity specialists that biology may be the next field, after cybersecurity, that will require urgent protection as new models become more powerful and capable of handling complex scientific tasks.
Much of the work happening in the chemical and biological space is considered dual use, meaning that the same information used to create a bioweapon can also be used to develop an antidote or a vaccine. According to Dawn Bloxwich, whose team reviews and investigates suspicious activity across Google DeepMind's AI models, if society wants to see improvements in health, it must also accept the possibility of these tools being misused.
These concerns intensified after Anthropic's Mythos model triggered alarm this year over AI's cyber capabilities, and after OpenAI's models autonomously hacked into AI group Hugging Face in July. Experts anticipate a similar biological disruption in the near future, potentially involving a commercial system or a real societal-scale bioweapon incident. AI could be misused by terrorist groups, lone-wolf attackers or state actors to design biological weapons, create viruses or deploy existing harmful pathogens.
Several people working on biological risks pointed to Aum Shinrikyo, a cult that conducted a terror attack in Tokyo in 1995, releasing sarin nerve gas on three commuter rail lines during morning rush hour and killing 13 people while injuring thousands. The concern is that AI tools could lower the barriers for such groups to conduct similar attacks, and there is increasing evidence of groups already using AI to assist terrorism.
In July, research by the University of Cambridge revealed how Boko Haram militants used several AI tools, including OpenAI's ChatGPT, Anthropic's Claude, Google's Gemini, SpaceXAI's Grok and the Chinese company DeepSeek, with the use going beyond digital propaganda into kinetic combat operations and weapons engineering. Safeguards implemented by AI developers include using software to detect and block suspicious queries or output specific safe responses, as well as trusted access programmes to vet users for access to advanced capabilities and conduct general monitoring of account behaviour.
AI companies also test their models' biological capabilities and potential risks, often working with external specialist evaluators such as SecureBio and Red Queen Bio. However, experts note that there are limits to what AI groups can do to prepare for potential misuse, since detecting a jump in cyber capabilities is relatively straightforward, while assessing biological capabilities typically requires costly and slow wet-lab experiments.
Running studies that demonstrate whether AI can help build a virus is often not possible or advisable because of clear biosafety concerns, and it is impossible to test biology end to end because no one can ethically test someone actually building a bioweapon. In response, Anthropic's leading model, Fable, will fall back to its previous version, Opus 4.8, for most requests related to biology and chemistry as it is considered safer for these domains, while OpenAI said it trains models to safely handle dual-use biological requests, continuously monitors for any signs of abuse and works closely with biology, biosecurity and government experts
AI Safety Bioweapons Biosecurity Anthropic Openai Google Deepmind Dual-Use Research AI Policy Bioterrorism Pandemic Prevention
Ireland Latest News, Ireland Headlines
Similar News: You can also read news stories similar to this one that we have collected from other news sources.
Pádraic Joyce Concludes Tenure as Galway Senior Football ManagerLegendary player and manager Pádraic Joyce has stepped down from his role leading the Galway senior football team after a period of significant success and stability.
Read more »
Aoife O’Brien Claims First Stage Win at Rás na mBan22-year-old Aoife O’Brien took the race lead after winning the 97km second stage of Rás na mBan in Gorey, leading an Irish clean sweep of the podium.
Read more »
The Fallen: The Magdalene Laundries and Ireland's Legacy of SilenceLouise Brangan's The Fallen: The Magdalene Laundries and Ireland's Legacy of Silence, which draws on archives and survivors' testimonies in The Fallen to reveal the dark history of the laundries and Ireland's legacy of silence, is one of eight shortlisted titles selected from almost 400 international submissions for the 2026 Cundill History Prize. Administered by McGill University, the Cundill History Prize is the world's leading award for history writing. The $75,000 prize is awarded annually to the book that demonstrates excellence across the prize's guiding criteria: craft, communication and consequence. The two runners-up each receive $10,000. From the world's first Jewish ghetto to Indigenous enslavement in North America, from Chinese life under American racial law to the reshaping of modern South Asia and the global life of an everyday commodity, the shortlist highlights books that uncover forgotten voices and speak to our current context of conflict, nationalism and globalised commodity chains. The shortlist also illuminates untold histories such as the key role of Christianity in the personality cult at the heart of North Korea's dictatorship, recovers overlooked voices like those of the women incarcerated in Ireland's Magdalene Laundries, and shines new light on familiar subjects, including that of the American constitution. Together, they showcase the range and excellence of contemporary history writing. Chaired by Emily Greenwood, James F. Rothenberg Professor of Classics at Harvard University, the 2026 jury includes leading historians and writers Ned Blackhawk, Fergal Keane and Barbara Weinstein. Greenwood said: "I feel a great sense of excitement and conviction about this shortlist. Each book on this list invites readers into an absorbing prose world and leaves them with a different reckoning of history."
Read more »
The Irish Roots of American Labor DayExplore how Irish immigrants, with their history of agrarian activism, played a pivotal role in shaping the American labor movement, leading to the establishment of Labor Day as a national holiday.
Read more »