AI x Bio Hackathon
Build the biosecurity infrastructure that doesn't exist yet.
AI can already evade many of biosecurity's best defenses. DNA foundation models like Evo2 have been jailbroken to generate sequences with over 90% similarity to SARS-CoV-2 and HIV-1. Current synthesis screening matches against known pathogens but AI-designed proteins can preserve dangerous function while looking nothing like anything in the database.
Benchtop DNA synthesizers are approaching virus-length capability with no mandatory screening. Metagenomic surveillance is generating real data but has no AI layer to process it. The Biosecurity Modernization and Innovation Act of 2026 has identified gaps that haven't been filled.
The infrastructure to handle what's coming is not built. This hackathon is about building pieces of it.
Co-organized by BlueDot Impact, Apart Research, and Cambridge Biosecurity Hub.
Prizes
On-site winners:
3k to 1st.
1.5k to 2nd.
500 to 3rd.
Concurrent online prizes: $4,000 total. $1,000 first place. Track prizes from CBAI and Sentinel Bio ($500/$500 each) plus fast-track to CBAI Fellowship. Top projects invited to the Apart Fellowship.
Who this is for:
No biosecurity experience required. Check out our Biosecurity course to get up to speed. Teams will form at the start, and we'll help people find effective cross-discipline teams.
We think this is especially relevant if you work in or study ML, security research, software engineering, or policy - and want to spend a weekend on something that matters!
Schedule
Saturday morning: threat landscape briefing, team formation!
Rest of Saturday: build.
Sunday: build then demos and feedback and after that chill and hangout!
Food and drinks provided.
What you get:
A working project - some prototype, tool, or technical contribution that addresses a gap! On Sunday afternoon you'll show what you built and get feedback from people who work in biosecurity and AI safety. Strong projects will be connected to funding, our incubator, or other paths to keep driving forward your work.
A way into working on humanity's most critical vulnerability. If your starting point is "I want to work on AI x bio but don't know where to start" — you'll leave with a scoped problem, a project you can point to, and a network you didn't have on Friday.
Tracks
DNA Screening and Synthesis Controls (sponsored by CBAI)
Build screening tools that work against AI-designed sequences. Current systems rely on homology matching, so comparing orders to known pathogens. That fails when AI can design novel proteins with no sequence similarity to anything on record. Develop functional risk assessment approaches that go beyond pattern matching.
Benchtop Synthesizer Security (sponsored by Sentinel Bio)
No mandatory on-device screening. No tamper-proof logging. No split-order detection across machines. The security architecture for benchtop synthesizers has been sketched but barely built. Turn specs into working prototypes: cryptographic screening, digital certificates for air-gapped devices, centralized fragment-assembly detection.
Pandemic Early Warning
Metagenomic sequencing can detect any pathogen, including novel ones. The CASPER network is generating deep wastewater sequencing data nationwide. What's missing is stuff like anomaly detection at scale, protein structure prediction for functional risk, and integration pipelines connecting sequencing data to public health response.
Biosecurity Tools and Risk Assessment
The catch-all track for projects that address a gap but don't fit the other three. Functional risk scoring for novel sequences. Automated assessment of AI-designed proteins. Cross-institutional threat intelligence. If RAND is right that tacit knowledge no longer protects us, the field needs tools it doesn't yet have.