A concise, fact-based update for security and risk professionals. Covering today’s highest-priority developments.

🔐 Core Security Intelligence

1) OpenText launches AI-centric cybersecurity suite across identity, data, and applications

What’s new:
OpenText announced a major expansion of its cybersecurity portfolio (Release CE 25.4), integrating AI-driven capabilities across identity, data, applications, and forensics. Components include Core Threat Detection & Response, Core Identity Foundation, and Application Security Aviator.
Source: Help Net Security – OpenText expands cybersecurity capabilities with AI integration

Why it matters:
As organizations move AI from experimentation to production, foundational domains; identity, data, and application security, must evolve accordingly. OpenText’s move reflects the shift from “AI as innovation” to “AI as infrastructure.”

Defenses:

• Inventory AI assets and governance zones. Apply the same identity, data, and app controls to AI workflows as you do to human users.
• Treat AI systems as privileged entities. Enforce least privilege, continuous auditing, and full telemetry of agent actions.
• Embed lifecycle observability. Ensure every model has rollback, monitoring, and data lineage tracking.

Expert Insight:
Organizations that govern AI as infrastructure—not as experiment—will define the next phase of enterprise resilience. This is a maturity signal for AI risk management.

2) Palo Alto Networks unveils AI-driven platform with human-in-the-loop security agents

What’s new:
Palo Alto Networks introduced Prisma AIRS 2.0 and Cortex AgentiX, designed to secure AI models and applications throughout the lifecycle. The platforms include continuous red-teaming, real-time agent monitoring, and human oversight mechanisms.
Source: Reuters – Palo Alto Networks launches AI-driven security offerings

Why it matters:
Security is now being re-architected around AI. Agentic systems multiply decision points and complexity. Incorporating human-in-the-loop oversight helps prevent runaway automation or self-reinforcing model behavior.

Defenses:

• Demand transparency from AI vendors. Confirm how autonomy, oversight, and rollback are handled in AI tools.
• Catalog AI systems as critical assets. Include AI applications in your risk register and ensure change management applies to models.
• Train the SOC for agentic patterns. Monitor bursts of automated actions or “looping” behaviors typical of AI-driven operations.

Expert Insight:
AI is now both a defense asset and an attack vector. Palo Alto’s shift underscores the need to secure the “AI within security.”

3) “Vibe hacking” — AI-fueled cybercrime on the rise

What’s new:
Cybersecurity experts warn of a new wave of “vibe hacking”, where attackers use AI productivity tools to scale extortion, phishing, and fraud campaigns. The term describes threat actors leveraging LLMs and automation to manipulate trust and emotional cues at scale.
Source: The Economic Times – ‘Vibe hacking’: A new type of cybercrime is on the rise

Why it matters:
AI-driven content automation drastically lowers the skill barrier for social engineering. It’s no longer about tricking a single user—it’s about manipulating thousands simultaneously with algorithmically optimized messaging.

Defenses:

• Constrain automation-enabled tools. Review which internal AI tools have outbound email or social connectivity.
• Deploy behavioral analytics. Detect campaign-scale changes like mass message similarity or repeated phrase patterns.
• Educate on AI-aided deception. Teach users to question realistic tone and empathy cues crafted by AI models.

Expert Insight:
“Vibe hacking” illustrates how emotion and automation intersect. Traditional phishing defenses may fail when every message feels authentic.

4) Cisco and NVIDIA debut AI-ready network platform with embedded security

What’s new:
Cisco and NVIDIA announced an AI-native network platform with integrated compute and built-in security controls. The design, part of Cisco’s Secure AI Factory initiative, aims to optimize data flow for training and inference while protecting workloads in transit.
Source: UC Today – Cisco and NVIDIA launch AI network platform

Why it matters:
AI workloads now stress-test networks as much as compute. Embedding telemetry and segmentation into data fabrics is crucial to prevent cross-workload data exposure.

Expert Insight:
Defenders must extend visibility beyond endpoints and apps to the network backbone that feeds AI. Network observability and encryption are now AI risk controls.

5) Sublime raises $150 million to expand AI-powered email threat detection

What’s new:
Email-security startup Sublime Security raised $150 million to scale its AI-native platform, which identifies malicious content and “prompt injection” within email-based workflows.
Source: The Wall Street Journal – Sublime raises $150 million for AI-powered email security

Why it matters:
Traditional email filters miss AI-manipulated text and image payloads. Platforms trained on contextual intent rather than keywords are redefining how enterprises handle inbound threats.

Expert Insight:
The next generation of email defense is contextual—not lexical. AI-assisted detection can adapt faster than rule-based gateways, but it must be paired with privacy-by-design and model integrity checks.

⚠️ Updates / Follow-ups

No significant follow-ups to previously reported stories.

Summary Table

Categories: Cybersecurity News

Tags: Agentic AI, AI Governance, AI Security, Network Infrastructure, Phishing