OpenText says its Fortify tools now combine deterministic code scanning with AI analysis and remediation, aiming to help teams review AI-assisted code without handing security decisions to a black box. The practical idea is straightforward: let established rules filter findings first, then use AI where context and suggested fixes can add value.
OpenText Watch analysis
What happened
In a post published on 20 November, OpenText describes three Fortify capabilities: Agentic Analyzer, Aviator and Fortify skills. The company says Agentic Analyzer can plan, analyse with a chosen model and validate results, including removing duplicates and false positives before sending findings to Fortify. It supports more than 46 programming languages and lets teams set a maximum cost per scan.
OpenText says Aviator classifies findings as exploitable, false positive or suspicious, with explanations and proposed code changes. Those changes can be reviewed in pull requests. Fortify skills let AI assistants work with the platform, including generating compliance reports and checking whether vulnerable dependencies are reachable in an application.
Why it matters
AI can produce code quickly, but reviewing and fixing it still takes engineering time. OpenText’s proposed division of labour puts repeatable checks first and reserves AI for harder contextual analysis and remediation. That could make an AI-assisted workflow more manageable, particularly when teams need a record of what was flagged and why.
The details come from a vendor’s account of its own products and demonstrations. They show what OpenText says the tools do, not independent proof that they catch every vulnerability or reliably produce safe fixes.
Our read
The strongest idea here is the workflow, not the promise that AI will tidy up security by itself. Deterministic checks, explanations, audit history and a visible pull request give developers something to inspect. That is a more useful arrangement than asking an agent to fix code and hoping for the best.
Teams considering the tools should test them against their own code and review proposed changes before merging. Security automation earns trust one inspected fix at a time, which is less glamorous than “fully autonomous” and usually a better sign.
What to watch
- Whether teams can measure detection quality and false positives against their existing checks.
- How often proposed fixes need human changes before they are safe to merge.
- Whether scan-cost limits, audit records and model choices work as described in day-to-day use.
Discussion spark: Should AI-generated security fixes be allowed to merge automatically when checks pass, or should a person review every change?
Sources and evidence
- AI writes your code. Can your security keep up? (20 November 2026, 08:00 UTC)
Independent WittyWires tracker for public updates about OpenText. Not affiliated with or endorsed by OpenText; this is not an official account.