Crowdstrike Falcon now powers runtime defense in Nvidia’s LLMs

Generative AI adopted 18To 7% in the last two years. But at the same time, the enterprise focused security investments specifically on the risks of AI only 43%create a siAI gift in preparation as the rapid expansion of attack surfaces.

Much more 70% experienced a violation associated with at least one AI over the past year SANS Institute findings.

AI infrastructure was amazed by the state-funded attacks 218% Over year, as Crowdstritch’s 2025 global threat report reveals.

A hard reality is clear for cisos, security and SOC leaders. The placement of new AI models of new AI models is expanding the attacks of the institutions, and speaking in anonymity, VentureBeat’s traditional security tactics, strategy and technology tactics, strategy and technology tactics, strategy and technology, strategy and technologies protested. CyberSecure industry has reached a critical infection: Providing generative AI requires more than bolt-on tools; Requires a complete architectural turn

Fortunately, Crowdstrike also offers a new solution Integration, NVIDIA’s hybrid and very cloudy environments provide more than 100,000 enterprise-scale LLM.

The strategic response of Crowdstrika

Community CEO George Kurtz seized an urgent Last interview with VentureBeat: “Security cannot be closed; it must be internal. An important part of our strategy can always use security information as the main element of our main infrastructure.

“NVIDIA’s Nemo security provides a frame to assess the risk of AI. In an end interview with the work officer, Crowdstrike, VentureBeat.

Kurtz strengthened this strategic vision BoneRon“Generative AI helps to bend the time. With the internal, telemetry-controlled safety, we define threats at the speed of cars, probably determine and neutralize threats in the machine.”

Bernard, “Crowdstrike EU-native CyberCurity, AI-Native Cyberecure, the development of the AI ​​is how the development is based on the life of the AI.”

Crowdstrike divides FALCON security directly to NVIDIA’s AI infrastructure

Falcon provides direct NVIDIA’s LLM SECURITY, Crowdstrike, which actually provides time protection of the threats in which threats actually formed. Inside the AI ​​pipeline.

“AI is not an independent initiative.

By taking an installed approach, it allows you to constantly scan contains insensitive, poisoned databases, incorrect configurations, and unauthorized shadows continuously.

These are the factors that are approximately affected by these 64% of businesses. During the time of working hours, protecting and neutralizing daily signals, models, models of models, models, modification and hidden data to detect and neutralize hazards

Bernard, Falcon’s unique distinction announced his unique difference.

Bernard, Falcon’s critical processing period made the advantage: “LLMS expand the attack on the attack, and the risks are extensive telemetry and platform on the planet and platform, and the platform on-threatening exploration and platform allows the attacks to stop the attacks before it happens.”

‘Shadow AI’ risk, previous byod ‘wild wild west’ periods

“Shadow AI is one of the biggest and most of the time today,” said Bernard. The shadow AI is most common and often remained out of focus – risks in the enterprise environment. Security teams often do not know where the models are working, whose being founded or configured – traditional software management has completely passed.

The lack of appearance creates a real risk, especially sensitive information AI systems are made or obtained. Falcon Cloud Security opens this secret activity throughout the environment, makes it visible and effective. You can apply politics and reduce the risk after you have visibility. If you fly blindly, “said Bernard.

Crowdstrike President Michael Sentonas clearly noted the strategic advantages in A Previous VentureBeat Interview“In the coordination of attackers, in the endpoint and telemetry coordination, directly closes the Falcon’s integration into the AI ​​pipeline, it dramatically closes these gaps, and the real-time image of the Cisman and the attacks are closed sharply.” ⁸

Given another approach to generative AI security, develop an attractive new plan for Cisos, which is ai to identify and contains and contains. However, he also emphasizes the need for a serious evaluation: CISOS should not have a direct infrastructure of safety, and inspecting the accurate adaptation of organizations with different architectures, risk exposure and strategic security purposes.

In general, the EU users and technical decision-makers are working in the programs, Chatgpt, Microsoft Copilot, Anthropic Claus, Anthropic Clos, Anthropic Clod, or Organization, AI tools with a very different risks, 2000 and 2010 in the early period of 2000 and 2010 “Byod” Adoption of unsecured and unspecified smartphones at work.

However, the adoption curve of gene AI models between users is very steep and technology is developing faster than many players, more security increases the mine area.

From reagent to real time: Why safe security issues for the generative AI

If the traditional AI security means that trust in foreign scanning and placement interventions are the most important of enterprises and protection that are sensitive to accurate end points and protection surfaces.

Crowdstrika’s integration into the witness cloud security Integration of NVIDIA in Universal LLM, this dynamics, direct protection directly to the EI to continue to the Ettenthirt.

Bernard explained how Falcon has reduced the risks before the placement of AI-SPM: “Sahin is researching for unauthorized models and policy violations.”

Falcon is directly included in the NVIDIA’s AI infrastructure, for example, comprehensive model security, tracking and audit ability, an excellent and automated part of an automated part, an internal and automated part of a labor-intensive work.

Integration with Crowdstritch with NVIDIA is a tool for cisos and enterprise degree Gen AI security

The generative AI is rapidly strained by traditional perimeter-based security methods expands attack surfaces.

Typical threats, including submitted injections, data leaks and model poisoning, require more deeply visibility and higher accuracy and higher accuracy and control. The integration of Crowdstrika with NVIDIA’s LLM infrastructure is noteworthy for the architectural approach to solving these security gaps.

Cisos, security leaders and security controls for the verses they serve, offer material operating benefits, including the following, including AI life:

• In the inner zero-confidence scale: Automated security policy eliminates hand efforts, consistently applies zero reliable protection along each AI model.
• Proactive sensitivity reduction: Prior to completing the risks of determining and neutralizing risks, the attackers significantly reduce the windows of the opportunity.
• Continuous work time intelligence: In real time, the discovery managed by telemetry, emergency injection, model poisoning and unauthorized information, and blocks the threats as an exclidant of information.

Bernard stressed the need to operate a more integrative approach to the security of Generative AI security. “Especially sensitive or owned risks, especially in subtle vision and placement, we are making better visibility during operation.

The generative AI is not just a distinctive, but the foundation of the enterprise infrastructure, the installed security is no longer optional. Crowdstrike and Nvidia’s integration does not add only protection; Redefines how the AI ​​systems have already been established to resist the development of the trader.