Generate Biomedicines' AI-designed GB-0669 completes phase 1

AI system

System: Generate Biomedicines generative protein design platform (AI-designed monoclonal antibody GB-0669)
Model: Not assessed
Release: Not assessed
Application developer: Generate Biomedicines
Role: Not assessed
Configuration: Not assessed
Inputs and outputs: Not assessed
Autonomy: Not assessed
Workflow: A generative protein-design platform aimed at the conserved S2 stem helix. A randomized, placebo-controlled single-ascending-dose trial in healthy adults. Live-virus neutralization assays on participants' serum.
Human oversight: Not assessed
Relation to AI: The original design study describes machine-learning optimization of a known antibody using a structure-conditioned generative model and experimentally trained sequence-to-function models. The clinical study evaluates the resulting antibody’s safety and laboratory activity. It does not quantify AI’s advantage over an alternative discovery process or establish a clinical benefit to patients.
AI contribution: The original design study describes machine-learning optimization of a known antibody using a structure-conditioned generative model and experimentally trained sequence-to-function models. The clinical study evaluates the resulting antibody’s safety and laboratory activity. It does not quantify AI’s advantage over an alternative discovery process or establish a clinical benefit to patients.

Outcome identifiers: Not assessed
Benefit: dose-dependent serum live-virus neutralization occurred at 600 and 1,200 mg
Category: Improved Knowledge Discovery
Status: Not assessed
Beneficiary groups: People. Organizations. Society.
Beneficiaries: Not assessed
Reach: Not assessed
Distribution and access: Not assessed
Duration: Not assessed
Expected?: Yes
Costs and harms: Not assessed

Studies: Not assessed
Study design: Not assessed
Measurement: Not assessed
Setting and sample: Not assessed
Measures: dose-dependent serum live-virus neutralization occurred at 600 and 1,200 mg
Comparison: Not assessed
Results: dose-dependent serum live-virus neutralization occurred at 600 and 1,200 mg
Additional value from AI: Not assessed
Limitations: Not assessed
Denominators: Not assessed
Materials for replication: Not assessed

Participants: Developer: Generate Biomedicines. Operator: Generate Biomedicines with contract research organization PPD (phase 1 site).
Jurisdictions: Benefit: Orlando, Florida, United States (phase 1 trial). Developer: United States. Operator: United States.
Ownership and control: Not assessed
Dependencies: Not assessed
Cross-border connections: Not assessed
Geopolitical implications: Not assessed

Conditions for replication: Healthy adult volunteers, single intravenous ascending doses from 100 to 2,400 mg. Live-virus neutralization assay against circulating sarbecoviruses. Antibody manufactured as in the trial.
Replication attempts: Not assessed
Later findings: 14 July 2026. The results were subsequently published in the Journal of Infectious Diseases. The authors propose further clinical evaluation; prevention or treatment of infection has not been demonstrated by this study.
Availability: Not assessed
Institutional decision: Not assessed
Grounds: Not assessed
Action and result: Not assessed
Success criterion: Not assessed

Source numbers: AIBD-0047-source-0001–0003
Citations: Numbered references below. [1] [2] [3]
Language: Not assessed
Source relationships: Developer [1] [2] [3]
Original evidence: The original preprints are available. The published full text and two supplements were inaccessible, and the accounts differ on trial allocation. [1] [2] [3]
Access and versions: Not assessed
Supported findings: Not assessed
Evidence links: Not assessed
Evidence strength: Not assessed
Missing evidence: Not assessed

Assessed benefits: People. Organizations. Society.
Depth: Not assessed
Reasons: People: The trial enrolled healthy volunteers and measured safety, pharmacokinetics and serum neutralization; no person obtained a therapeutic benefit. Organizations: The source reports a completed phase 1 and a recommendation to proceed to phase 2, not a realized organizational outcome. Society: No population-level benefit is demonstrated; the antibody has not been shown to prevent or treat infection in patients.
Observed or intended: Not assessed
Rubric: Not assessed
Assessor and date: Not assessed
Alternative assessments: Not assessed
Fill: The makers' claim

Kind: Not assessed
Related entries: Not assessed
Field: Antiviral biologics / clinical development. Healthy adult volunteers.
Purpose: Use generative machine learning to design a therapeutic antibody against a conserved, poorly immunogenic region of the SARS-CoV-2 spike protein.
Dates: Source published: 8 October 2025. Updated: 24 October 2026.
Stage: Not assessed
Version: SHA-256: e988731b6fe0…

Peer-reviewed study Developer

Borriello F, Koh GCKW, Troconiz IF, et al.. Randomized, Double-Blind, Placebo-Controlled First-in-Human Trial of a First-in-Class AI-Designed Monoclonal Antibody (GB-0669) Against the Conserved SARS-CoV-2 Spike S2 Stem Helix. 2026-07-14. Journal of Infectious Diseases. Published abstract reviewed alongside the full October 8, 2025 author preprint, doi:10.1101/2025.10.07.25337449.

makers
Preprint Developer

Marden GC, Schmitt KP, Zimmerman MI, et al.. Functional and structural characterization of a combination of pan-sarbecovirus antibodies with potent antiviral activity. 2025-03-31. Original machine-learning design and preclinical study, cited as reference 14 by the clinical preprint.

makers
Dataset Developer

Generate Biomedicines; ClinicalTrials.gov. NCT07050511: GB-0669 in healthy subjects. 2025-07-03. Sponsor-submitted registration posted after trial completion; no posted results. Structured allocation field conflicts with the manuscript’s randomized description.

makers