Skip to content
Apart Research
Apart Research
Contact us
Sprint projectSep 15, 2025India

Molecules Under Watch: Multi-Modal AI Driven Threat Emergence Detection for Biosecurity

Subramanyam Sahoo · Team One Man Army

Submitted to CBRN AI Risks Research Sprint. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: Molecules Under Watch: Multi-Modal AI Driven Threat Emergence Detection for Biosecurity

Open in a new tab (opens in new tab)
Code (opens in new tab)
Share

This study presents a comprehensive multi-modal pipeline for assessing biosecurity risks in chemical compounds, integrating real and synthetic datasets from public repositories such as ChEMBL, PubChem, and USPTO patents. The system leverages molecular descriptors extracted via RDKit, contextual embeddings from the Qwen-2.5 language reasoning model, and unsupervised anomaly detection using Isolation Forest to compute a novel Threat Emergence Detection (TED) score. This score quantifies dual-use potential, synthesis feasibility, and novelty, enabling scalable threat triage. We evaluate the pipeline on hybrid datasets, demonstrating robust differentiation between real pharmaceutical compounds and synthetic benchmarks. Our approach advances AI-driven CBRN (Chemical, Biological, Radiological, Nuclear) safety by providing interpretable risk metrics and constitutional oversight, with implications for regulatory compliance and dual-use research governance.

Reviews

Judging this Sprint?

Review “Molecules Under Watch: Multi-Modal AI Driven Threat Emergence Detection for Biosecurity”

Your public critique appears on the project page without your name. Reviews of strong projects are shared with funders, as you choose below.

Not shown on the project page.

Shown on the project page, without your name.

Only the Apart team reads this, and funders if you agree below.

Decline to score

Only when the quality of the writing makes the work impossible to follow. Say why in your private critique; the project page shows only that a judge declined, not why.

Does the project meaningfully address the theme of CBRN risks and AI misuse? Does the project clearly connect to a CBRN-related threat (chemical, biological, radiological, nuclear)? Does it build on or challenge existing literature or prior work in the space? Does it offer a novel framing, tool, or evaluation that could inform future research or policy? Is it grounded in real-world relevance rather than hypothetical misuse?

Does this project advance the field of AI safety in a clear and valuable way? Does it help identify or mitigate risks specific to powerful AI systems? How well does the project engage with the unique challenges of AI alignment, misuse, or system evaluation? Could the output contribute to scalable safety mechanisms or evaluation pipelines? Does it generalize to broader dual-use or frontier AI concerns?

Is the project thoughtful, usable, and technically sound? Is the technical implementation solid and well-scoped for a research sprint? Is the code or method clearly documented and reproducible? Does the project demonstrate sound methodology, clear assumptions, and an honest treatment of limitations? Would the tool or prototype be useful for future research, governance, or evaluation work?

Sharing your name *

When we share a strong project with funders, your name, scores and public critique go with it, confidentially; this choice is only about whether your name is public. If you think this project deserves funding, please also endorse it: that carries the most weight.

Share my private critique with funders *

See our Privacy Policy.

  1. This project is ambitious and makes a valuable contribution by tackling the challenge of detecting potentially dual-use chemical compounds. The integration of molecular descriptors with a composite “TED score” is a creative and promising approach, and the methodology is laid out clearly.

    Some design choices, such as assigning fixed 0.8/0.2 threat probabilities or weighting TED score components at 0.4/0.3/0.3, come across as somewhat arbitrary. These would benefit from either empirical justification or sensitivity analyses.

    Overall, this is a very strong project: impressive in scope, well-executed, and directly relevant to AIxCBRN risk. This could evolve into a genuinely useful tool for practical biosecurity screening.

  2. The project is critical in the AI-CBRN interface, the scoring framework is novel and implementable. My only note to the author is that the binary threat probability assignment (0.8 vs 0.2) is overly simplistic, and you could potentially miss chemical structural signals.

Cite this project

@misc{sahoo2025molecules,
  title = {{Molecules Under Watch: Multi-Modal AI Driven Threat Emergence Detection for Biosecurity}},
  author = {Subramanyam Sahoo},
  year = {2025},
  month = sep,
  note = {Submitted to CBRN AI Risks Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/molecules-under-watch-multi-modal-ai-driven-threat-emergence-detection-for-biosecurity-7x9q}},
  url = {https://apartresearch.com/sprints/projects/molecules-under-watch-multi-modal-ai-driven-threat-emergence-detection-for-biosecurity-7x9q}
}
Browse all projects

Build something like this at the next Sprint

Swarm Alignment Hackathon · Nov 6 - 8, 2026