The new fear in the US: AI could assist in creating biological weapons

Artificial intelligence has already proven it can write code, assist in research, and solve complex problems. Now, its advanced capabilities in the field of biology are raising a different kind of concern in the United States: could it potentially facilitate the creation of dangerous pathogens in the future and even assist in carrying out a biological attack?

According to a report published on Monday in the Washington Post, the White House is working to strengthen US defenses against biological attacks, while concern grows among officials and experts regarding the evolving biological capabilities of AI systems. However, these efforts come after cuts and organizational changes at the beginning of President Donald Trump's second term significantly reduced personnel and expertise in the field of biosecurity within the government.

Team reduced — and the technological threat advances

According to three former senior officials who spoke with the newspaper, at the end of President Joe Biden's term, a team of up to 30 professionals focused on biosecurity operated in the White House. Its role was, among other things, to coordinate between federal agencies dealing with biological threats and to provide scientific advice to senior government officials. After Donald Trump's re-entry into the White House, the team was quickly reduced as part of cuts and organizational changes. According to the sources, at certain times there was no one in the White House whose sole role was to deal with biosecurity.

At the same time, the administration canceled a 2023 executive order on artificial intelligence, which included a call to strengthen biological defenses, and cast doubt on the policy from the Biden era intended to require companies to perform checks on customers seeking to purchase genetic components that could be used to create dangerous biological agents. Donald Trump ordered the formulation of a revised version of the policy by August 2025, but according to the Washington Post, more than a year later it has still not been published. A White House official told the newspaper that the administration is working to strengthen the previous policy, which he said was technically limited and insufficient.

The concern: advanced biological capabilities of AI

In the background lies the rapid advancement of artificial intelligence systems in biological research. According to the report, officials in the AI industry have long warned about the possibility that the tools could be used to carry out a biological attack. The fear is that language models and dedicated tools for biological design could in the future make it easier to perform complex laboratory operations or assist in designing new pathogens, including those that existing monitoring systems would find difficult to detect. However, the Washington Post emphasizes that among scientists there is still disagreement regarding the severity of the risk, the time when it might materialize, and the right way to prepare for it. The report also points to a recent scientific development: this month, results of a study were published in the journal Science in which artificial intelligence was used to engineer new viruses. The viruses were designed to infect bacteria and not humans, so it cannot be concluded from this that AI is already capable of producing a lethal human virus; however, according to the article, the experiment demonstrated the progress of capabilities in the field.

Not just a threat: AI may also be used for defense

The same technology that raises concern may also be used as part of the defense. According to the report, work is underway at the Pentagon aimed at harnessing artificial intelligence to accelerate the development of defenses against future pandemics and to strengthen deterrence against biological warfare. Part of the work includes analyzing billions of possible changes in antibodies to examine how it will be possible to respond to new forms of the coronavirus. Also involved in the efforts is a former health official who led Operation Warp Speed — the operation established during Donald Trump's first term to accelerate the development of vaccines for coronavirus. AI companies are also cooperating with the administration. According to a person familiar with the matter, a series of briefings was recently held between OpenAI and the White House and about half a dozen government agencies on the subject of biological risks. The company is also working to grant government officials access to a dedicated version of GPT for biosecurity, known as Rosalind.

The question that remains open

The challenge arising from the report is twofold: AI capabilities in the field of biology are advancing rapidly, but precisely during this period, the American government apparatus that is supposed to deal with biological threats has undergone significant cuts and changes. The White House, for its part, claims that its commitment to dealing with AI risks and biosecurity is clear and ongoing and that it has been developing policy on the subject for more than a year. On the other hand, experts and former senior officials who spoke with the Washington Post warn that the loss of personnel and expertise could harm the ability to formulate long-term policy. Thus, while the question of how much AI will really facilitate the creation of biological weapons is still a matter of scientific debate, in the United States they are already taking the possibility seriously — and are trying to rebuild the defense system in a race against the developing technology.