SecureBio

📋Page Status

Page Type:ContentStyle Guide →Standard knowledge base article

Quality:65 (Good)⚠️

Importance:48 (Reference)

Last edited:2026-01-31 (1 day ago)

Words:1.4k

Structure:

📊 3📈 0🔗 7📚 13•10%Score: 13/15

Issues (2):

QualityRated 65 but structure suggests 87 (underrated by 22 points)
Links1 link could use <R> components

Quick Assessment

Dimension	Assessment	Evidence
Focus Area	AI-bio risks + pathogen surveillance	Two divisions: AI team (capability evals) and NAO (wastewater monitoring)¹
Funding	Well-funded by EA sources	≈$9.4M from Open Philanthropy across multiple grants²³⁴
Policy Influence	Growing	Virology Capabilities Test adopted by major AI labs; NIST AI Safety Consortium member⁵
Team	≈25 staff	Experienced leadership from MIT, biosecurity, and operations backgrounds⁶
Key Concern	Information hazard balance	Biosecurity research inherently involves dual-use knowledge⁷

Overview

SecureBio is a 501(c)(3) nonprofit founded in 2022 by Kevin Esvelt, an MIT professor who invented CRISPR-based gene drive technology.¹ The organization works to protect against catastrophic pandemics—both natural and engineered—through a three-part framework: Delay dangerous capability proliferation, Detect novel pathogens early, and Defend through physical and institutional interventions.⁸

The organization operates two main divisions. The AI & Biotechnology Risks team, led by Seth Donoughe, evaluates how AI systems might accelerate biological threats and develops benchmarks for AI labs. The Nucleic Acid Observatory (NAO), led by Jeff Kaufman, pioneers wastewater-based metagenomic surveillance to detect novel pathogens before they spread widely.⁶

SecureBio occupies a distinctive position in the biosecurity landscape by explicitly focusing on catastrophic and existential biological risks rather than routine public health threats. The organization also bridges the AI safety and biosecurity communities, recognizing that advances in AI could significantly lower barriers to creating engineered pathogens.⁵

History

Founding and Early Development (2022-2024)

SecureBio was founded by Kevin Esvelt, who brought credibility from his work on gene drive technology and his security-minded approach to biotechnology. Esvelt had previously argued that the “security mindset was not sufficiently present in most of the bioengineering industry.”⁹

The organization quickly attracted funding from Open Philanthropy, receiving multiple grants:

$4,000,000 for general support over three years²
$3,430,000 for the Nucleic Acid Observatory program³
$1,420,937 for biosecurity research over three years⁴
$570,000 for pathogen early warning systems¹⁰

Leadership Transition (2024-Present)

In September 2024, Kevin Esvelt stepped back from SecureBio’s Board to focus on his MIT professorship and academic research, though he remains involved as a co-founder.¹¹ Dr. Benjamin Mueller, formerly COO, became Executive Director and Chairman of the Board.

The board was restructured with new members bringing diverse expertise:

Christine Parthemore - CEO of Council on Strategic Risks, national security expert on WMD/CBRN threats
Michael Specter - Staff writer at The New Yorker, science and public health journalist
Liv Boeree - Science communicator and philanthropist¹¹

Core Framework: Delay, Detect, Defend

Delay

The Delay pillar focuses on slowing the spread of dangerous biological capabilities to potential bad actors. Key initiatives include:

SecureDNA: An international initiative providing free DNA synthesis screening software to prevent dangerous pathogen creation. The system can detect hazardous genetic sequences down to 30 base pairs.¹¹

AI Capability Restrictions: Working with AI labs to implement appropriate access controls for models that could assist with biological weapon development.⁵

Detect

The Detect pillar centers on early warning systems for novel biological threats:

Nucleic Acid Observatory (NAO): Uses untargeted metagenomic sequencing of wastewater to identify new and unknown pathogens—including those with extended incubation periods that evade symptom-based surveillance. As of 2025, NAO monitors wastewater from “more than a dozen sewersheds as well as three military facilities.”¹¹

The approach is deliberately pathogen-agnostic, meaning it can detect threats regardless of whether they are natural, accidentally released, or deliberately engineered.¹

Defend

The Defend pillar develops physical and institutional measures to prevent infection and maintain societal resilience:

Far-UVC Research: Previously investigated germicidal ultraviolet light as a passive defense against airborne pathogens (currently inactive).⁸

Pandemic-Proof PPE: Advocated for improved personal protective equipment standards.⁸

AI & Biotechnology Risks Program

SecureBio’s AI team has become increasingly prominent as concerns about AI-enabled bioterrorism have grown.

Virology Capabilities Test

The team developed the Virology Capabilities Test (VCT), a benchmark measuring AI models’ ability to assist with complex virology tasks. A 2025 study using the VCT found that OpenAI’s o3 model “outperformed 94 percent of expert virologists” on troubleshooting complex lab protocols, demonstrating an “urgent need for thoughtful access controls.”¹²

The VCT is now used by major AI laboratories for pre-release safety evaluations.¹¹

NIST Engagement

SecureBio participates in the NIST US AI Safety Consortium and submitted formal recommendations including:⁵

Ensuring the AI Risk Management Framework addresses biosecurity risks from foundation models
Developing evaluations for CBRN risks including static benchmarks, model-graded evaluations, and task-based evaluations
Assessing biological design tools (BDTs) that could lower barriers for non-experts

Collaboration with AI Labs

SecureBio collaborates with frontier AI companies including Anthropic to build evaluation tools and mitigation strategies. Notably, former SecureBio Research Scientist Anjali Gopal, who co-led the AI project, moved to Anthropic’s technical staff.⁶

Team and Organization

Current Leadership

Leadership Team

Ben Mueller

Executive Director & Board Chair

Seth Donoughe

Director of AI

Jeff Kaufman

Director of NAO

Alvaro Morales

Director of Operations

Team Structure

Division	Head	Team Size	Focus
AI & Biotechnology Risks	Seth Donoughe	≈12	AI capability evaluations, policy recommendations
Nucleic Acid Observatory	Jeff Kaufman	≈15	Wastewater surveillance, metagenomic sequencing
Operations	Alvaro Morales	≈3	Administration, finance, HR

Key researchers include Jasper Götting (Head of Research, AI team), Will Bradshaw (Head of Computational Programs, NAO), and James Kremer (Head of Laboratory Science, NAO).⁶

Notable Departures

Kevin Esvelt - Stepped back September 2024 to focus on MIT professorship¹¹
Anjali Gopal - Former AI Project Co-Lead, now at Anthropic⁶

Funding

Major Grants

Funder	Amount	Purpose	Year
Open Philanthropy	$4,000,000	General support (3 years)	2023
Open Philanthropy	$3,430,000	Nucleic Acid Observatory	2023
Open Philanthropy	$1,420,937	Biosecurity research (3 years)	2022
Open Philanthropy	$570,000	Pathogen early warning	2022
Total identified	≈$9,420,937

Operating Costs

The AI-Bio evaluation project alone costs approximately $700,000 per 6 months, covering team leads, policy scientists, contractors, and research assistants.¹³ As of early 2024, SecureBio was actively seeking additional funding to sustain operations.¹³

Criticisms and Concerns

Information Hazard Dilemma

SecureBio’s work inherently involves dual-use knowledge—understanding how to defend against biological threats requires understanding the threats themselves. The broader biosecurity community remains divided on how to handle such information hazards.⁷

Some experts argue that emphasizing certain risks could “fuel threats rather than mitigate them”—citing how Al Qaeda’s bioweapons efforts reportedly began after the United States publicly highlighted biosecurity vulnerabilities.⁷

SecureBio has been praised for being “mindful of information hazards” in its approach,⁹ but the tension between transparency and security remains unresolved in the field.

Scope Limitations

Evaluators have noted that SecureBio’s expertise “concentrates on biological threats, not financial systems or weapons infrastructure vulnerabilities.”¹³ The organization is not positioned to address all catastrophic risks.

Funding Dependence

Like many EA-aligned organizations, SecureBio relies heavily on Open Philanthropy funding. This concentration creates sustainability risks and potential influence concerns, though no specific issues have been raised.

Key Uncertainties

Key Questions (5)

Can DNA synthesis screening scale globally before advanced AI models make biological weapon creation significantly easier?
Will wastewater surveillance provide sufficient warning time to prevent catastrophic pandemic spread?
How should AI labs balance capability advancement with biosecurity concerns?
Can information hazard frameworks be developed that enable defensive research without enabling attacks?
Will SecureBio's AI capability evaluations keep pace with rapidly advancing AI systems?

Relationship to AI Safety

SecureBio represents an important bridge between the biosecurity and AI safety communities. The organization’s work on AI capability evaluations directly addresses concerns about AI misuse risks, specifically the potential for AI to lower barriers to biological weapon creation.

The fact that SecureBio staff have moved to AI labs (Anjali Gopal to Anthropic) and that AI labs use SecureBio’s evaluation tools suggests meaningful knowledge transfer between these communities.

Sources

SecureBio Official Website - Organization overview and mission ↩ ↩² ↩³
Open Philanthropy - SecureBio General Support - $4M grant ↩ ↩²
Open Philanthropy - SecureBio NAO - $3.43M grant ↩ ↩²
Open Philanthropy - Biosecurity Research - $1.42M grant ↩ ↩²
SecureBio NIST RFI Submission - AI policy recommendations ↩ ↩² ↩³ ↩⁴
SecureBio Team Page - Staff and organizational structure ↩ ↩² ↩³ ↩⁴ ↩⁵
Bulletin of the Atomic Scientists - Biosecurity Community Divided - Information hazard debates ↩ ↩² ↩³
Founders Pledge - SecureBio Assessment - Impact evaluation ↩ ↩² ↩³
EA Forum - SecureBio Notes from SoGive - Independent assessment ↩ ↩²
Open Philanthropy - Pathogen Early Warning - $570K grant ↩
Letter from the Executive Director - SecureBio Substack - Leadership transition and current status ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶
Virology Capabilities Test Paper - AI virology benchmark results ↩
SoGive Notes on SecureBio - Funding and operational details ↩ ↩² ↩³