newsfilter.io
Interview

AI-engineered diseases are coming. Here's the plan to stop them. | Andrew Snyder-Beattie

  • Open Philanthropy anticipates dispersing hundreds of millions of additional dollars in the future, with a "crash program" to implement its four-pillar biosecurity strategy (personal protective equipment, protected buildings, detection, and medical countermeasures) achievable within approximately two and a half years.
  • The organization expects its four-pillar strategy to reduce extinction risks from biological catastrophes by at least 50%, potentially significantly more, with a probability estimate of a 1% to 3% chance of a survival-threatening biological catastrophe occurring within the next 25 to 50 years.
  • Andrew Snyder Beattie projects a 50% probability of a pandemic as severe or worse than COVID occurring within the next 25 years and warns that low-tech countermeasures like elastomeric respirators are critical to "buy time" while advanced medical solutions develop, noting current US stockpiles are largely expired.
  • Future biological threats are expected to be more dangerous than 1980s Soviet weapons, particularly regarding AI systems that might gamble on releasing pathogens to eliminate humanity or use the threat as a negotiating position; the risk of AI accelerating biological weapon creation is estimated at least 10%.
  • Mirror bacteria are assessed as having a greater than 10% chance of being catastrophic if released tomorrow but a less than 80% chance of causing total extinction, with the primary risk being the gradual death of plants and ecosystems due to their ability to evade immune systems and phage predation.
  • Response strategies for mirror bacteria could rely on respiratory protection, as masks might provide a two-order-of-magnitude reduction sufficient to sustain life, alongside environmental sterilization using chemicals like propylene glycol, triethylene glycol, and hypochlorous acid, which have sufficient US production capacity to cover industrial and residential spaces.
  • Current infrastructure for food security is robust, with the US holding enough stockpiled food for 18 months (extending to two years if post-harvest), and theoretical capacity to feed the population for 500 years using synthetic bacteria fed by natural gas and electricity.
  • Despite these defenses, the outlook identifies significant resource gaps, with fewer than 100 full-time personnel working on worst-case biological scenarios, zero staff dedicated to glycol vapor supply chains, and only three personnel on the personal protective equipment project.
  • Geopolitical and state actors, specifically Russia and North Korea, are expected to maintain active offensive biological weapons programs, while bureaucratic interests may sustain such programs even when strategically redundant, and the US Department of Defense has already ordered elastomeric respirators.
  • The window of vulnerability exists between the ability to create dangerous biological weapons and the scaling of defensive technology, with norms against mirror bacteria creation currently reducing extinction risk from a 1-2% chance to approximately 0.25%.
  • Individual recommendations include acquiring good elastomeric respirators and stockpiling shelf-stable food for a three-month social distancing period, as human survival in such scenarios may depend on immediate, cooperative action rather than waiting for government intervention.