Summary: Cohere announced on October 15, 2026 a proposed agreement to acquire AISLE, a company whose AI systems find and help fix previously unknown software vulnerabilities. The deal would put autonomous security research inside Cohere’s sovereign AI platform, with AISLE’s founders leading a new Cohere Cyber business unit. For builders shipping AI into regulated or critical environments, it signals that model vendors are starting to treat software security as part of the deployment stack, not a separate purchase.
What AISLE actually does
AISLE, founded in 2024, builds autonomous systems that discover, validate, and help remediate unknown vulnerabilities in production software. The track record from its first year out of stealth is concrete: more than 500 CVE-assigned zero-days found across over 100 codebases, including all 12 vulnerabilities in OpenSSL’s January security release and 18 of the 24 CVEs in curl’s latest release, per Cohere’s announcement. Maintainers independently reviewed those findings.
I covered one of those results earlier this year — AISLE’s OpenSSL release analysis — and the interesting part then was the same as now: a foundation model wrapped in specialized tooling (large-scale code analysis, rigorous validation, automated remediation) can outperform what either the model or a human team does alone.
Why Cohere wants it
Cohere frames the deal as part of its sovereign AI push, following acquisitions of Reliant AI and Aleph Alpha. The stated logic: enterprises moving AI into production face requirements around data protection, governance, compliance, and the risks of more autonomous systems, and meeting them needs a full-stack combination of models, cybersecurity, and secure deployment rather than powerful models alone.
There’s also a research angle worth noting. Cohere says difficult, verifiable security problems could serve as evaluations that help improve future frontier models. Security bugs come with ground truth — either the exploit works or it doesn’t — which makes them a cleaner training and evaluation signal than most enterprise tasks.
What changes for deployments
If the acquisition closes, AISLE’s three founders take leadership roles: Ondrej Vlcek as SVP of the new Cohere Cyber business, Jaya Baloo as Cohere’s Chief Information Security Officer, and Stanislav Fort as VP of AI Security within core model research. Cohere also plans a substantial long-term investment in Prague as its global center for AI cybersecurity.
For customers, Cohere lists expected outcomes including threat detection across enterprise and public sector deployments, stronger governance and observability for AI agents, and support for private cloud, on-premises, and air-gapped environments. Concretely, that means the vendor you buy models from would also be auditing the software those systems run on — a bundling of concerns that used to sit with separate security vendors.
The builder takeaway
This fits a pattern: sovereign AI platforms competing on where and how safely models run, not just on benchmark scores. Cohere has been assembling exactly this stack — its North platform targets deployment control for utilities, and AISLE adds the security layer underneath.
Two caveats. First, this is a letter of intent, not a closed deal, so treat the customer benefits as expectations rather than shipped features. Second, autonomous vulnerability discovery cuts both ways — the same capability accelerates attackers. The practical move for teams evaluating sovereign AI vendors now is to ask which security guarantees are contractual today versus promised post-integration.