By Tom Sittler
Read and comment on the Google Document version of the post here.
This is part of a series of posts about our progress in the first five days of the Oxford Prioritisation Project.
Laurie Pycroft: “Gaining an overview of Biorisk”
From Laurie’s research report:
The top three sources of existential and global catastrophic risk over the early to mid 21st century are likely to be thermonuclear war, artificial superintelligences, and engineered biological agents. Of these, I intend to briefly address biological technology risk, focusing particularly on the risks associated with CRISPR and related genetic engineering technologies, excluding the risk from gene drives (which will be addressed separately by Dominik. CRISPR (clustered regularly interspaced short palindromic repeats) is a novel biotechnology capable of exerting extremely fine control over genetic systems through use of bacterial proteins, and doing so at a fraction of the expense and effort that more traditional systems incur.
One major limitation I encountered is that, while genetic engineering risk has been discussed in some form for several decades, the novelty of CRISPR means that it is not as widely discussed. The seminal publications that launched CRISPR into widespread awareness among the scientific community were published in February 2013 [1,2] and broader understanding of the potential risks posed by the technology did not arise until later. Thus, only work published in the last four years is likely to be of very direct utility. That said, much earlier work may still be of use if it addresses general problems with genetic biorisk instead of specific technological concerns.
Risks specific to, or highly exacerbated by, CRISPR:
● Low barriers to entry. Simple DIY CRISPR kits are crowdfunded as science experiments for children and the equipment required for a more comprehensive laboratory can be purchased easily from eBay. The principles of CRISPR are part of the cell biology syllabus at most major universities, and the practical application is widely taught to graduate students in relevant fields. Together, this means that malign actors (e.g. terrorists) would not find it difficult to establish a functioning CRISPR lab. It also means that many legitimate labs worldwide can work on the topic, which may be good from the perspective of development of countermeasures, but it also increases risk of accidental release of engineered pathogens.
● Easy design of pathogens. Gain-of-function experiments are almost trivially easy to conduct by the standards of bioscience. This means that legitimate labs that do not have appropriate biosafety level 3 or 4 facilities can easily perform experiments that would be risky even with such facilities. Particularly concerningly, gRNA sequences for CRISPR experiments can be ordered online from many suppliers. Oversight of such orders appears to be limited and, while some suppliers do claim to screen for harmful sequences, some do not and, even if all did, it would be an intractable problem to identify all possible harmful sequences.
● Gene drives (see Dominik’s work)