Identity becomes the control plane for enterprise AI security

The stolen credentials are responsible for 80 % of the violations of the institutions. Each major security seller has experienced the same conclusion: the identity is now an artificial intelligence control plane. The scope alone requires this transformation. Institutions that run 100,000 employees will deal with more than a million identity when artificial intelligence agents enter production.

The traditional structure of identity access management (IAM) cannot expand the scope of the proliferation of AI. It has been built for thousands of human users, not millions of independent factors that quickly work with the machine with human levels. The industry response is the most important security transformation since the adoption of cloud computing.

Near -based authentication replaces device symbols

Bluetooth Low Energy (BLE) is now used between devices and laptops to prove physical proximity. Besides encryption identities and biological measurements, this creates the approval of four factors without the need for organs symbols.

CICO Di -CIVE explains this innovation on a large scale. The verification of their proximity provides the approval resistant to the Ble -based proximity in conjunction with vital verification. This ability, which was revealed in Cisco Live 2025, is a fundamental transformation of the authentication structure.

The Entra ID of Microsoft deals with 10,000 artificial intelligence agents in one experimental programs with 8 billion authentication per day. “The traditional guide services of independent systems that operate is not so quickly,” says Alex Simps, CVP of identity in Microsoft.

Ping IDENTITY’s Davinci Orvictration Platform pays more. The system processes more than a billion authentication event daily, as artificial intelligence agents represent 60 % of traffic. Complete all verification in less than 200 milliliters while maintaining the encryption guide.

Behavioral analyzes pick up the factors at risk in real time

Crowdstrike treats artificial intelligence agents like any other identity threat. Their Falcon platform creates the behavioral foundation lines for each agent within 24 hours. Deviations lead to automatic containment within seconds.

“When Amnesty International’s agent suddenly arrives at the systems outside his fixed style, we deal with them in analogy with the adoption of the exposed employee.” The statute tracks 15 billion of Amnesty International events daily through customer environments.

This speed is important. Crowdstrike Global Threat Document Documentations that opponents achieve initial access in less than 10 minutes. They move sideways across 15 systems during the first hour. Artificial intelligence factors working with identities at risk of exaggeration of this damage significantly.

Identity prevents catastrophic failures

Average 89 different identity stores through cloud and vitality systems, according to Gartner. This retail creates blind spots that opponents use daily. Reform applies network principles to identity infrastructure.

Access to the advanced server from OKTA repetition, loading and automatic failure through identity providers. When the initial approval fails, secondary systems are activated within 50 milliseconds. This becomes obligatory when artificial intelligence agents carry out thousands of operations per second.

“The identity is safe, you give the agents access to real systems, real data and customer data,” said Tod McKinnon, CEO of OKTA, in OKTANE 2024.

Zero scales confidence for the spread of the agent

Palo Alto Networks’ Cortex Xsiam completely abandons the surrounding defense. The platform assumes the ongoing settlement. Each Amnesty International is subject to verification before each procedure, not only in the initial approval.

Mike Rimmer, Ciso Square in Ivanti, strengthened Zero Trust in a recent interview with Venturbeat: “It works on the principle of“ Never Trust, always achieved. ”By adopting a zero confidence structure, institutions can ensure that only users and authenticated devices can access data and sensitive applications.”

CICO’s global ZTNA extends this model to artificial intelligence agents. The platform expands Zero Trust beyond humans and the Internet of Things to include self -self -intelligence systems, providing a wide discovery and a large -scale mandate.

Mechanism books immediately respond to abnormal cases of identity. When harmful programs lead to authentication violations, XSIAM eliminates access and forensic analysis without human intervention. The zero classification response becomes this operating basis.

Jay Chaudhry, CEO of ZSCaler, identified the basic weakness of Zenith Live 2025: “Network protocols are designed to allow reliable devices to freely communicate. AI weapon of this old structure on a large scale.

Ztna’s global work frameworks allow the publication of one million workers

Architectural requirements are clear. Universal Zero Trust Network Access (ZTNA) provides four capabilities necessary for artificial intelligence environments.

The Cisco app shows the desired scale. The ZTNA global platform performs automatic discovery checks every 60 seconds, which leads to the indexing of new artificial intelligence spread and ear groups. This eliminates the blind points targeted by attackers. CISCO’s authoritarian mandate imposes the lowest accounting limits through policy engines that address 100,000 decisions per second.

Comprehensive auditing paths pick up every work agent for criminal investigation. Safety teams that use platforms such as CISCO can rebuild accidents across millions of reactions. The original support for standards such as the form of the context of the context of the model guarantees the ability to operate with the development of the ecosystem.

IVANTI approach complements these capabilities through behavior analyzes. Darren Joyce, SVP Product Management at IVANTI, confirms: “Safety tools can be acting on behalf of the Acting Point of Acts to analyze huge amounts of data to detect abnormal cases and predicting potential threats faster and more accurate than any human analyst. These tools provide a clear vision through devices, users and networks, and to determine the potentially possible security.”

Security architecture puts artificial intelligence in CISCO towards industry

Cisco Ai Secure Factory puts them as the first NVIDIA Silicon in the NVIDIA reference architecture. By combining post -quarter encryption and new devices, CISCO builds infrastructure to protect from threats that have not yet existed. Enterprise Takeway: Artificial Intelligence Security is not optional; It is an architect.

In Cisco Live 2025, the company revealed a comprehensive identity and safety strategy of Amnesty International that treats each layer of stack:

advertisement	The basic problem / strategic value has been solved	Technical details	Availability
Hybrid protection wall (including Hypershield)	Distributor, original fabric security; Safety from the ocean to the network tissue	Empf; enforcement; Speeding devices	New Protection Walls: October 2025
Protection	The “correction” gap is closed for 45 days for 3 days “with rapid weakness at the nucleus level	Patching in actual time without restarting	OS Nexus: September 2025
Splunk: Swallowing the free protection wall record	Reduces SIEM costs up to 80 %; CISCO adoption stimulates	Swallowing an unlimited record of Cisco protection walls	August 2025
Splunk: AI’s observation susceptibility	It provides a decisive vision in the performance of the artificial intelligence	It monitors the use of the graphics processing unit and the performance of the form	September 2025
I am the duo	It develops from MFA to the fully security IAM platform	User Guide, SSO, ID Order	Available now
The duo: proximity verification	Provides authenticity resistant to deception without devices	Near -based proximity, biometric verification	Part of the new duo
The duo: elasticity of identity	It addresses the risks of critical nitrogen interruption	Repetition, loading and automatic failure	In development
Cisco Universal Ztna	It expands zero confidence to humans, Internet of Things/OT and AI agents	Automatic discovery, delegated delegation	Continuous development
Open source safety form	democracy from artificial intelligence; 8B parameters match the performance of 70B	It works on the central processing unit. 5 b training of security symbols	Available (embracing face)
Ai Defense and Nvidia Partnership	It secures the artificial intelligence development pipeline	Nvidia Nim Microservices Optimization	Available now
Posttum Security	Future resistance against quantum attacks	MacSec and IPSEC	New devices (June 2025)
ID intelligence	Continuous behavioral monitoring	Male anomalies detection	Part of the security cloud
Safe arrival	The VPN and ZTNA capabilities converge	The edge of the safe access service that was delivered by the cloud	Available now

Cooperation across the seller is accelerating

The Cloud Security Alliance Zero Trust Advancement Center now includes every main safety seller. This unprecedented cooperation provides unified safety policies across platforms.

“The security sellers must unify the common threats,” George Courts, CEO of Croldstrike, confirmed during a discussion of a modern platform strategy. “The approach centered on data wins the extent of the development of the two numbers and threats.”

“Security is a prerequisite for adopting artificial intelligence. If people do not trust the system, they will not use it.”

The organizational challenge remains. Robert Gratoli, Ivante Information Manager, determines the decisive barrier: “The alignment of CISO and CIO will be very important in 2025. This cooperation is necessary if we want to protect modern companies effectively. Executive officials need to unify resources – budgets, employees, data and technology – to enhance post -security in the organization.”

Identity account

When CISCO, OKTA, ZSCaler, Palo Alto and Crowdstrike networks independently reach identical conclusions about identity structure, it is confirmation, not a coincidence.

Identity infrastructure determines the results of security. Organizations face two options: a professional identity as a control or acceptance plane as inevitable. The gap between the speed of spreading artificial intelligence and the maturity of the seat of identity is narrowing daily.

You cannot wait three procedures. Review the identity of each AI agent and its ears within 30 days. Spread the continuous verification of all non -human identities immediately. Establishing ID safety operations 24/7 to prevent opponents from exploiting gaps.

The consensus of the seller sends a clear significant signal. Identity has become an artificial intelligence control plane. Institutions that fail to adapt 2025 will eliminate violations instead of innovation.