Coding Agents Are Changing the Biosecurity Risk Landscape | GovAI
As coding agents grow more capable, they could help actors circumvent safeguards by modifying or building harmful biological AI tools. Biosecurity efforts need to adapt.
Coding Agents Are Changing the Biosecurity Risk Landscape | GovAI About Research Opportunities Team Analysis Alumni Updates Donate Coding Agents Are Changing the Biosecurity Risk Landscape As coding agents grow more capable, they could help actors circumvent safeguards by modifying or building harmful biological AI tools. Biosecurity efforts need to adapt. By Luca Righetti, Kamile Lukosiute , and James Black Research Posts April 20, 2026 GovAI research blog posts represent the views of their authors rather than the views of the organization. You can read about GovAI’s governance structure
saved by
related reading
- Paving the way for agents in biology \ Anthropicanthropic.com
- Modifying LLM Beliefs with Synthetic Document Finetuningalignment.anthropic.com
- A biosecurity playbook for AI companies — EA Forumforum.effectivealtruism.org
- How to Solve AI Biosecuritythecounterfactual.substack.com
- Malleable software in the age of LLMsgeoffreylitt.com
- Designing agentic loopssimonwillison.net
- The lethal trifecta for AI agents: private data, untrusted content, and external communicationsimonwillison.net
- [2312.06942] AI Control: Improving Safety Despite Intentional Subversionarxiv.org
- Self-Fulfilling Misalignment Data Might Be Poisoning Our AI Modelsturntrout.com
- Coding Models Are Doing Too Much | whnrehiew.github.io
- Self-exfiltration is a key dangerous capabilityaligned.substack.com
- Cognitive Security as an AI Safety Cause Area — LessWronglesswrong.com