Skip to content
Cohere — AISLE — cubic — AI — News

Cohere Completes AISLE Acquisition and Buys cubic to Expand Cybersecurity and Code Review

The deals add autonomous vulnerability research and secure software development to Cohere’s sovereign AI platform.

Table of Contents

Cohere has completed its acquisition of AISLE and acquired AI code-review company cubic, adding autonomous vulnerability research and software development tools to its enterprise AI platform. AISLE forms the foundation of a new cybersecurity business, Cohere Cyber, while cubic’s team will establish a secure-coding business within the company.

The two additions extend Cohere’s work beyond supplying models and the infrastructure to run them. They bring tools for finding security flaws, reviewing software changes, and fixing code into a platform aimed at enterprises, governments, and critical-infrastructure operators that need control over their data and computing environments.

Cohere counts AISLE and cubic among four strategic M&A transactions completed this year, following Aleph Alpha and Reliant AI. The company is assembling capabilities across frontier models, healthcare and biopharma, cybersecurity, and software development.

AISLE Becomes the Foundation of Cohere Cyber

AISLE builds autonomous systems to discover, validate, and help remediate previously unknown vulnerabilities in production software. Its approach combines foundation models with specialized security expertise, large-scale code analysis, validation, and automated remediation.

That work goes beyond generating a plausible warning about a piece of code. Validation establishes whether a suspected flaw is a real vulnerability, giving developers and security teams a concrete issue to investigate and fix.

When Cohere outlined the proposed AISLE acquisition in October, it reported that AISLE had discovered more than 500 previously unknown vulnerabilities assigned CVE identifiers across more than 100 codebases in its first year out of stealth. CVEs are standardized identifiers used to track publicly disclosed security vulnerabilities. The findings were independently reviewed by the affected software’s maintainers, Cohere said.

AISLE’s three co-founders will take roles across the commercial security business, Cohere’s internal defenses, and model research. Ondrej Vlcek will become senior vice president of Cohere Cyber and lead the new business. Jaya Baloo will become chief information security officer, overseeing security across Cohere. Stanislav Fort will become vice president of AI Security, joining core model research and training while leading AI security research.

Cohere aims to bring AISLE’s security work closer to frontier-model development, using difficult, verifiable security problems to help evaluate and improve future models. It also plans a substantial long-term investment in Prague, expanding AISLE’s existing presence into a global center for AI cybersecurity and autonomous security systems.

Extending AI Tools Inside Customer Security Boundaries

Both deals share a deployment requirement. Cohere’s sovereign AI strategy is built around letting organizations retain control of their data, infrastructure, and operations while using advanced models and applications. Supported environments include private clouds, customer-operated infrastructure, and air-gapped systems isolated from external networks.

Code review and vulnerability research analyze sensitive material, including proprietary source code and details of flaws in critical software. Cohere says cubic’s review agents were built to support on-premises and air-gapped deployments, extending that deployment approach into software engineering.

The acquisitions also complement Cohere’s existing inference infrastructure, the systems that run trained models to produce responses. Earlier this month, the company added third-party open models to Model Vault, its managed inference platform. Model Vault uses dedicated, single-tenant infrastructure, offers regional processing in the United States, European Union, and Canada, and retains neither prompts nor outputs by default.

Model hosting, security research, and code review address different parts of the same operational workflow.

Comments

Latest