Tenzai Expands Its AI Hacker to AI Applications

The record-breaking autonomous offensive safety firm extends its full-stack testing to incorporate AI methods, protecting net, API, and AI software layers in a single run.

NEW YORK CITY, NY / ACCESS Newswire / June 1, 2026 / Tenzai, the autonomous offensive safety firm ranked high 1% in opposition to 125,000 human hackers throughout six international platforms, immediately introduced the extension of its AI hacker to incorporate AI functions, enabling enterprise safety groups to check the entire assault floor of recent AI methods, from the net layer and APIs by means of to the AI software itself.

Right now’s announcement comes as AI functions enter manufacturing at a tempo that has outrun present safety infrastructure. Gartner projects that 40% of enterprise functions will embed task-specific AI brokers by the tip of 2026, up from lower than 5% in 2025. A big share of these functions are being constructed by AI coding brokers, with Tenzai’s own research discovering that each main AI coding device – Cursor, Claude Code, Codex, and Devin – shipped susceptible code when examined in opposition to an identical prompts and environments. Unbiased analysis corroborates the sample: 45% of AI-generated code samples introduce OWASP Top 10 vulnerabilities, and the results are compounding – March 2026 alone saw more CVEs directly attributed to AI-generated code than all of 2025 combined.

A Totally different Method to AI Software Safety

Tenzai’s strategy to AI software safety is completely different from the market norm. The vast majority of AI software safety tooling out there immediately was constructed from the mannequin outward: it checks whether or not the mannequin resists immediate injection, whether or not it leaks coaching information, or whether or not guardrails maintain. Tenzai’s strategy begins from the total assault floor and works inward.

That distinction issues due to what AI software vulnerabilities really seem like in follow. A immediate injection is usually the entry level, however the underlying vulnerability is extreme device authority, a lacking authorization examine, or a credential that propagates to a workflow that shouldn’t inherit it. These are traditional safety bugs made reachable by means of the AI layer. Discovering them requires an agent that understands each the AI habits and the infrastructure it operates in.

When Tenzai’s agent encounters an AI software, it doesn’t run a prompt-injection guidelines in opposition to the mannequin. It maps the appliance as a set of actors, directions, instruments, credentials, guardrails, state transitions, and endpoints and generates leads throughout all of them.

“Probably the most consequential vulnerabilities in AI functions don’t reside within the mannequin, they reside in what the model-driven brokers have entry to,” stated Pavel Gurvich, CEO and cofounder of Tenzai. “Increasing Tenzai’s AI hacker to AI functions is the pure subsequent step: our hacker already traces assault paths throughout the net layer, the APIs, and now the AI habits in the identical run, repeatedly, with each take a look at constructing on what the final one realized. The findings that matter are nearly all the time chains, so testing one layer in isolation, or testing as soon as 1 / 4, misses them.”

Watch a demo and skim extra about Tenzai’s AI software hacking right here: LINK

About Tenzai:

Tenzai is an AI-native cybersecurity firm constructing autonomous AI hackers to make sure enterprises ship unbreakable code. Its platform actively hacks, exploits, and helps repair vulnerabilities throughout enterprise software program – repeatedly and at scale. Based in 2025 by cybersecurity veterans Pavel Gurvich, Ariel Zeitlin, Ofri Ziv, Itamar Tal, and Aner Mazur, Tenzai has raised $75 million in seed funding from main traders together with Battery Ventures, Greylock Companions, Lux Capital, and Swish Ventures. Learn extra: http://www.tenzai.com

Media contact:
Itai Singer, TellNY
[email protected]

SOURCE: Tenzai