KR — ZDNet Korea

Oracle and OpenAI Team Up to Tackle AI Security Threats: "Find and Fix Vulnerabilities"

SecurityOctober 28, 20262 min read

Oracle and OpenAI are teaming up to help companies find, check, and fix security weak spots in their software. The new service uses OpenAI's AI security models together with Oracle's security testing tools.

What happened

At Oracle AI World 2026 in Las Vegas, Oracle CEO Clay Magouyrk and OpenAI CEO Sam Altman talked about how companies can handle AI security threats. Oracle launched the Oracle AI Security Mesh. An AI agent (a program that can carry out tasks on its own) checks possible attack paths and weak spots, then suggests fixes. Customers must approve any changes before they are made. The two companies also expanded their partnership to include ChatGPT links to Oracle business systems and an AI app-building tool.

Key facts

The event took place on May 27, 2026 (local time), at the Venetian Expo in Las Vegas.
The Security Mesh works on Oracle Cloud Infrastructure (OCI), other clouds, and on-premises systems (computers a company runs in its own building).
In a live demo, the system found an API (a way for programs to talk to each other) that could be used without a login.
Oracle plans to create a new AI consulting team based on OpenAI technology.

Why AI is changing security

AI can now find old bugs in software very quickly. Magouyrk said AI is not creating new security problems. Instead, it is helping people find weak spots that have built up over the last 50 years. Altman agreed that the only real answer is to find and fix these problems, and he said companies need a new kind of service that checks their main traffic all the time.

Why it matters

Many companies are starting to use AI agents, and each new agent can open new doors for attackers. This partnership shows that big tech firms see AI as both a risk and a tool for protection. If the system works well, companies could find and fix holes in their systems faster than before.

What to watch

Whether companies actually adopt the Security Mesh and how well it works outside demos.
How the AI agents handle approvals so that changes are safe.
How the new ChatGPT and Oracle business links are used in real work.
What the new AI consulting team offers to customers.

Glossary

Oracle AI Security Mesh: A new Oracle service that uses AI to find, test, and fix security weak spots in company systems.
OCI: Oracle Cloud Infrastructure, Oracle's cloud computing service where companies run software and store data.
API: A set of rules that lets different programs talk to each other and share data.

Creinity Responds to PHP Security Vulnerabilities, Highlights Java-Based Email Security Technology

Creinity, an AI-based email security and collaboration company, said its SpamBreaker inbound email security solution does not include a PHP runtime and is therefore not directly affected by the recently disclosed PHP vulnerabilities, which allow remote code execution without authentication. The company said its core components, including the mail processing engine and management console, are built on Java, and that SpamBreaker supports three-layer inspection along with SPF, DKIM, and DMARC sender authentication. Creinity also cited CC certification for SpamBreaker and GS certification for several other products.

SecurityArs TechnicaOct 28

Code-execution flaw in UEFI firmware threatens most servers on planet Earth

A memory-corruption flaw, tracked as CVE-2026-97631 and rated 9.5 out of 10, affects the TianoCore EDK II codebase used to build UEFI firmware for most server vendors and data centers. The bug lies in an ICMP component and can let attackers run malicious code before the OS and Secure Boot defenses load. Because AMI published a pull request with a fix before the coordinated disclosure date, TianoCore issued a public advisory last week, and the publicly available patch makes exploit development easier.

SecurityThe RegisterOct 28

EU privacy advocates accuse US data annotation firm of snooping on contractors

Privacy group noyb has filed five GDPR complaints in France, Italy, the Netherlands, and Austria against Scale AI, alleging it unlawfully tracks European contractors working on its Outlier AI platform. The complaints claim Scale collects device and activity data, including activity outside of Outlier work, and shares it with third parties such as LinkedIn, TikTok, and Facebook. Scale has not been quoted in this article responding to the allegations.

SecurityVnExpressOct 27

Vietnam "Must Prepare Now" for Quantum Security Risks

Experts warn that the "harvest now, decrypt later" threat, in which encrypted data is collected today and decrypted once quantum computers are powerful enough, is growing, and that Vietnam should prepare response plans early. Speakers at an event in Ho Chi Minh City during AEF 2026 said finance, banking, public services and critical digital infrastructure could face the risk of complete code-breaking, and that moving to post-quantum cryptography (PQC) is now urgent. They also noted that cryptographic migration is a lengthy process affecting the whole technology infrastructure, and that implementation readiness remains uneven across banks and businesses.