NEW YORK, Nov. 19, 2026 (GLOBE NEWSWIRE) -- Escape, the offensive security engineering platform, today launched Escape Continuous AI Pentesting, a new product that runs a pentest on what each software release changes. Security teams can trigger it on every merge, after every deploy, or on a daily, weekly, or monthly schedule.t It covers web applications, APIs, and MCP servers. In Escape's benchmark, runs scoped to a change finished 16x faster than a full assessment and caught 91% of the vulnerabilities each change introduced.
The window between a vulnerability’s disclosure and its exploitation has dropped from 32 days in 2022 to hours, according to Zero Day Clock. CISOs who have spoken with Escape report 10,000 to 20,000 code commits a week, and they don’t have the budget or time to grow the security teams that test that code. They also say they need to know within 24 hours whether a release exposed them, or it’s simply too late. An annual manual pentest can’t keep up with that, and most AI pentesting tools still run point-in-time assessments that are out of date by the next release.
"Most security teams still get a pentest once a year and spend the other 11 months hoping nothing they shipped in between opened a hole. That doesn't work when engineers push thousands of commits a week and attackers can exploit a new flaw within a day," said Tristan Kalos, CEO and co-founder of Escape. "Escape Continuous AI Pentesting tests each change as it ships, so teams know in minutes whether a release exposed them, and they have the logs to show auditors exactly what we tested."
How Escape Continuous AI Pentesting works
Escape's whitebox integration reads the commits since the last run on each watched repository. Then each run:
- Analyzes the changed code and reports the issues that change introduced.
- Maps the change to every route it reaches, up to 20 per run. A change to a middleware or an authentication helper can open risk on routes whose code never moved. Tools that only check the changed files miss that.
- Attacks those routes on the deployed application with the agents of Cascade Iguazu, Escape's multi-agent AI pentesting engine. When a route fronts a chatbot, an AI agent, or an MCP server, the run applies Escape's AI red teaming techniques.
- Replays every open issue against the live application in parallel. An issue closes only when its fix is verified.
Teams pick one trigger per profile. On every merge, the Escape GitHub App starts a run when someone pushes code to the watched branch, and teams don't have to add a workflow. After a deploy, the customer's CI calls Escape's API so the run tests what's live. Or teams can set a daily, weekly, or monthly schedule. Escape runs the tests on its own side, so builds never wait on them. Escape reads Go, Python, TypeScript, JavaScript, Java, Ruby, PHP, C#, Rust, C, C++, Kotlin, Swift, and Scala.
Fewer duplicate tickets, findings tuned to each organization
Security teams worry that testing more often means a bigger backlog. Escape keeps every open issue as a single issue across runs using a deduplication agent. Known issues come back as replays of the existing ticket, and Escape verifies fixes automatically.
Customers can also import past pentest reports, bug bounty findings, and other context documents. Escape turns them into attack scenarios specific to that organization. If five past reports were cross-site scripting (XSS), the run gets tailored XSS scenarios, and those scenarios apply across every profile in the organization, so one team's past finding sharpens testing on every team's apps. Each finding goes to its owner with remediation guidance, or a prompt ready for the AI model the developer uses.
"A company’s attack surface is growing faster than they can test it. AI-written code, new applications and agents with real permissions are all multiplying, and attackers are using the same models to move faster," said Antoine Carossio, CTO and co-founder of Escape. "By 2027, I expect AI-driven attackers to autonomously discover and exploit previously unknown complex vulnerabilities within minutes, without relying on public disclosures or human-written exploits."
Audit evidence for every change
On every run, Escape's reasoning logs record what it tested and why, and developers get the reproduction path for each finding. That matters for frameworks like PCI DSS v4.0.1, which requires penetration testing at least once a year and after any significant application change. Customers who need more can add human-made, pentest reports from OSCP-certified testers from the same platform, so they don't have to switch platforms for each type of test.
"Before opening our MCP server to customers, we wanted it tested the way AI-driven attackers would approach it. Escape's AI Pentesting fit into our pre-launch process and gave our dev team clear, actionable results we could act on quickly," said Jun Carreon, Principal Information Security Analyst at Arkose Labs.
Availability
Escape Continuous AI Pentesting is available now, directly and through Escape's channel partners. Learn more at escape.tech/product/continuous-pentesting.
About Escape
Escape is pioneering offensive security engineering, a new approach that replaces legacy scanners and manual processes with AI agents that discover, test, and remediate vulnerabilities directly in engineering workflows. Built for security teams that are 100x outnumbered by developers, Escape automates the full offensive security lifecycle so teams can multiply their impact without scaling headcount. Learn more at escape.tech.
Media contact
Sherri Walkenhorst
sherriw@connectmarketing.com
(801) 373-7888