SandboxAQ Advances AI Security, Drug Discovery and Battery Tech in Multi-Sector Push

SandboxAQ continued to showcase its multi-industry quantitative AI strategy this week, unveiling advances in cybersecurity, drug discovery and materials science. The company emphasized a human-centric approach to AI while positioning its platform as an enabling layer for enterprises in biopharma, energy and other regulated sectors.

In cybersecurity, SandboxAQ expanded its AQtive Guard offerings with a Public AI Security Index that rates large language models and MCP servers on jailbreaks, misuse, toxicity, security and robustness. The firm also highlighted enhancements to its Open Cryptography tool, which now prioritizes exploitable cryptographic risks in Docker images to reduce noise for security teams.

These security developments aim to help enterprises assess third-party AI models before deployment and strengthen cryptographic hygiene in modern software stacks. While immediate monetization details remain limited, the tools could embed SandboxAQ more deeply into AI risk management and DevSecOps workflows if widely adopted.

In life sciences, SandboxAQ spotlighted a preprint on using synthetic, physics-based data to improve AI models for drug discovery, expanding its SAIR dataset with about 80,000 free energy perturbation calculations. New SAIR-FEP and SAIR-OOD benchmarks are intended to better capture real-world and out-of-distribution scenarios for binding affinity prediction.

The research indicates that sequence-based models benefit from added physics descriptors and that filtering for high-confidence co-folded structures is key for structure-based models. Jointly training on synthetic and experimental data improved performance on public benchmarks, pointing to a scalable data-augmentation strategy for pharmaceutical R&D.

The company also promoted AQAffinity, an open-source model built on OpenFold3 in collaboration with NVIDIA to accelerate drug–target affinity prediction. By lowering time and compute requirements and fostering a developer community, SandboxAQ is seeking to entrench its tools as infrastructure for AI-driven drug discovery.

Beyond biopharma, SandboxAQ underscored its battery technology efforts, preparing to present its Large Quantitative Models for battery materials discovery and capacity fade prediction at The Battery Show Asia 2026. This extension into energy storage aligns the firm with electric vehicle and grid-scale applications where improved degradation modeling has clear economic relevance.

Leadership commentary at the Cures Conference reinforced a focus on using quantitative AI to rescue drugs that fail clinical trials due to adverse effects, targeting high-value late-stage development challenges. Overall, the week’s updates portray SandboxAQ as deepening its technical stack and market positioning across AI security, healthcare and materials science, laying groundwork for diversified enterprise partnerships over time.