Cohere, an AI company that sells language models to businesses, signed a letter of intent on October 15, 2026 to buy AISLE. The startup's AI finds security holes in software. Neither side disclosed a price. The deal matters because AI bug finding is turning into a business, and Cohere wants to sell it to companies and governments.
What AISLE does
AISLE uses AI to scan code for vulnerabilities, the mistakes attackers use to break into systems. It was founded in 2025 by Ondřej Vlček, former CEO of the Czech antivirus maker Avast, with co-founders from Google DeepMind and Anthropic, Reuters reports.
Its own figures are striking, though no outside group has checked them:
| AISLE claim | Figure |
|---|---|
| New bugs found in widely used software in its first year | More than 500 |
| Vulnerabilities found in OpenSSL, since fixed | 12 |
| Vulnerabilities found among those fixed in curl's latest release | 18 of 24 |
OpenSSL is a widely used encryption library that protects web traffic. Curl is a widely used tool for moving data over the internet. BetaKit reports the curl figure from AISLE's blog, and adds that models from Anthropic and OpenAI found no issues in that test.
What changes after the deal
The deal is not final. A letter of intent sets out the plan before a binding contract is signed.
BetaKit says all of AISLE's roughly 50 full-time employees will join Cohere, citing an AISLE spokesperson. AISLE's team will form a new unit called Cohere Cyber:
- Ondřej Vlček, AISLE's CEO, becomes senior vice-president of Cohere Cyber.
- Jaya Baloo, an AISLE co-founder, becomes Cohere's global chief information security officer.
- Stanislav Fort, AISLE's co-founder and CTO, becomes vice-president of AI security.
Cohere co-founder Ivan Zhang wrote on X that "there are meaningful gaps in solutions available for [cyber] defenders." Cohere's statement, quoted by Reuters, said that "increasingly capable AI systems are transforming both cybersecurity and the threats organizations face."
The two reports disagree on where AISLE is based. Reuters calls it San Francisco-based. BetaKit says it is headquartered in Prague, Czechia, where its engineering hub will become Cohere's Prague office.
The bigger Cohere plan
AISLE adds to Cohere's push into Europe. Cohere is also combining with Germany's Aleph Alpha, the maker of the Kolibri open-weight model. Reuters calls it a merger reported to be worth about $20 billion. BetaKit describes it as an acquisition that still needs regulatory approval.
Reuters says the AISLE deal is meant to strengthen security and threat detection for business and public-sector customers. Cohere sells private deployments that run on a customer's own systems, so the code being scanned does not have to leave the building.
What this means for developers
AI bug finding is moving from research into products you can buy. Anthropic already offers free scans for open-source projects, and now a second major model company is building a security unit around it.
If you maintain open-source code, expect more AI-found bug reports. AISLE's curl and OpenSSL results show these tools can find real flaws, so set up a clear way to receive and check security reports.
If you buy security tools, ask how a vendor's AI results are checked. The figures here come from AISLE itself. Ask any vendor for its false-positive rate, meaning how often it flags code that is actually safe, before you trust the headline numbers.
For Cohere customers, the private-deployment angle matters most. Code scanning that runs on your own servers avoids sending source code to an outside service.
This article was first published on Tech AI Wire.
Also available in
Deutsch · 日本語 · Français · Español · Português
Related on Tech AI Wire
- Aleph Alpha Kolibri 1: open German-English MoE model
- Anthropic Cyber Mission: free scans for open-source code
Sources
- Cohere to acquire Czech AI cybersecurity firm Aisle, set up new cyber division - BetaKit
- Cohere signs letter of intent to buy AI cybersecurity startup Aisle - Reuters via The Globe and Mail
Top comments (0)