Developments surrounding artificial intelligence (AI) have been rapid in recent years. Think of ChatGPT, but there are also various biodesign tools such as ProteinBERT or DeepCRISPR. With these tools you can design proteins with certain therapeutic properties much faster and in a more targeted manner. The possibilities are enormous and AI offers many opportunities, but there are also risks.
The scenario: Designing toxic proteins with AI Biodesign tool
In your organization, research is being conducted into proteins that are involved in the signal transmission of cells of the immune system. The goal is to design new proteins that have a blocking effect with minimal toxicity. A postdoc recently received his PhD on this subject and is working with AI biodesign tools to design proteins. In the AI tool you can set all kinds of parameters, including the degree of toxicity. In a mad mood, the researcher suddenly decided not to design proteins with low toxicity, but rather structures with high toxicity. In no time, the AI biodesign tool identifies a number of proteins, both known and unknown, that are highly toxic. The postdoc is enthusiastic and tells colleagues, but also at parties, what he has discovered.
Risks and perspective for action
If you work in a biotechnology organization, the knowledge in your organization, for example about the toxicity of AI-generated proteins, may also be of interest to others who may have malicious intentions. Researchers often do not realize the potential of AI biodesign tools and that they can also be used for other purposes. Awareness of dual-use applications of such tools among researchers and responsible use of the technology is important to prevent misuse. Restricted access through passwords on the software or screening employees who have access to the AI biodesign tool can help reduce the risk of misuse. It is also important to consider where further safety steps can be taken, especially when the digital design of proteins is converted into the actual synthesis of the proteins.
Relevant links:
- Urbina, F., Lentzos, F., Invernizzi, C. et al. Dual use of artificial-intelligence-powered drug discovery. Nat Mach Intell 4, 189–191 (2022). [ignore]https://doi.org/10.1038/s42256-022-00465-9[/ignore]
- WHO - Emerging technologies and scientific innovations: a global public health perspective, 2023. [ignore]https://www.who.int/publications/i/item/9789240073876[/ignore]
- NTI - The Convergence of Artificial Intelligence and the Life Sciences [ignore]https://www.nti.org/analysis/articles/the-convergence-of-artificial-intelligence-and-the-life-sciences/[/ignore]
- Welcome Trust - Unlocking the potential of AI in Drug Discovery - Current status, barriers and future opportunities, 2023. [ignore]https://cms.wellcome.org/sites/default/files/2023-06/unlocking-the-potential-of-AI-in-drug-discovery_report.pdf[/ignore]
Sharing Your Scenario
If you would like to discuss how to handle this scenario with the Biosecurity Office or if you have encountered a biosecurity scenario that you would like to share with colleagues in the field, please feel free to contact us. The Biosecurity Office can be reached at [ignore]biosecurity@rivm.nl[/ignore].