Crowdstrike Falcon now powers runtime defense in Nvidia’s LLMs

Entrepreneial intelligence has increased by 187 % over the past two years. But at the same time, institution security investments have grown specifically on the risks of artificial intelligence Only 43 %, create SIA big gap in preparation as the surfaces of the artificial intelligence attack expand quickly.

More than 70 % of the institutions witnessed at least one violation of AI last year alone, with obstetric models now the primary goal, according to the results of the last SANS Institute.

The state’s attacks on the infrastructure of Amnesty International have increased by 218 % on an annual basis, and the Global Threat Report for 2025 reveals.

For CISO, security leaders and soc, the harsh reality is clear. The spread of the models of the new Amnesty International on a large scale expands the surfaces of the attack to its institutions, and it has told Ciso, who speaks on the condition that its Venturebeat identity is not disclosed that security and traditional technologies are facing a challenge to keep up with. The cybersecurity industry has reached a critical turning point: The Instems I intelligence requires more than the bolt tools; It requires a complete architectural shift

Fortunately, Croldstrike also introduces a new solution: on June 11 at the GTC Paris event in NVIDIA, the security company announced that it had guaranteed the security of the Falcon Cloud directly within the NVIDIA Universal LLM NIM. Integration provides more than 100,000 LLM circulations across the foundation via NVIDIA and multi -missile NVIDIAs.

Crowdstrike strategy response

George Courts, CEO of Crowdstrike, has the urgency in a recent interview with Venturebeat: “Security cannot be installed; it must be essential. An important part of our strategy has always been to take advantage of safety data as a major element in our basic infrastructure. You cannot secure artificial intelligence without clear data in the deepest layers.”

“NEMO safety in NVIDIA provides a working framework to assess the risks of artificial intelligence. The intelligence of the threat in Crowdstrike enhances this framework by enabling security and operations teams to build handrails about the emerging AI tactics-helps to evaluate the models that actually occur, where it actually occurs, where it actually occurs, where it occurs in anything. The chief business official, the chief business official,, Croldstrike, in a recent interview with Venturebeat.

KURTZ has strengthened this strategic vision to Barron, saying clearly: “The Introduction I intelligence helps us to bend time. With the built -in security, which is moved by remote measurement, we define threats and neutralize them quickly the machine, and most likely stops violations six times of traditional methods.”

Bernard stressed the importance, saying: “The crowds are a pioneer in the original cybersecurity, and we define how artificial intelligence is secured through the life development cycle. This last collaboration with NVIDIA brings our leadership to the forefront of the cloud, where LLMS is spread, runs, and lacks.

Crowdstrike Memorial Security Falcon is directly into artificial intelligence infrastructure in NVIDIA

By including the security of Falcon Cloud directly in the microscopic services of LLM NIM’s NIM, Croldstrike provides operating time protection where threats appear actually: inside the artificial intelligence pipeline itself.

“Artificial intelligence is not a stand -alone initiative – it has become included in all parts of the organization. Unlike many cloud security sellers who link the capabilities of artificial intelligence, we have built Amnesty International Security directly on the Falcon platform. This allows us to provide uniform protection across the cloud, and the points of end – which is very important as the attackers are increasingly moving, and not targeting a single surface.”

By following an embedded approach, Crowdstrike enables the Fallon to clear the constantly developed artificial intelligence models before publication, proactive discovery, poisoned data groups, poor formation, and unauthorized shade.

In combination these are factors that affect approximately 64 % of institutions. During the operating time, Falcon is enhancing the artificial intelligence that Crowdstrike, which is trained daily on trillion signals, to quickly reveal and neutralize advanced threats, including rapid injection, typical tampering, and filtering secret data.

Bernard highlighted clearly the unique Falcon Distinguishing during an interview with Venturebeat, saying: “What distinguishes us is simple: We secure the entire life intelligence cycle. With our integration in Nvidia LLM NIM, we give customers the ability to protect models before publishing them and while they are operated – while protecting the operating time that is presented through the same light agent that protects them.

Bernard has explained the critical time for operating time for Valcon, with an emphasis on: “LLMS works to expand the institution’s attack quickly, and the risks are already real. From immediate injection to API abuse, we have seen how sensitive can leak data without traditional breach.

“Shadow AI” risk re -mindy

Bernard warned: “Shadow Ai is one of the biggest proportions that are often ignored – today.” Shadow AI is one of the most common risks – often ignored – in institutions environments. Security teams often do not know where the models work, or who builds them, or how to configure them – completely transcends traditional software governance.

The lack of vision creates real risks, especially given that sensitive AI systems are trained or accessible. Falcon Cloud Security reveals this hidden activity across environments, which makes it visible and implemented. Once you get this vision, you can apply policy and reduce risks. “Without it, you are flying blind,” Bernard says.

“The attackers constantly control their methods, the exploitation of gaps in identity, the end point, and the coordination of remote measurement. The Falcon integration is direct The right.

Follow -up to a more inclusive approach in the safety of the IQ of Trucist is a convincing new scheme for CISO, which faces the challenges of identifying advanced artificial intelligence quickly. However, it also emphasizes the necessity of strict evaluation: CISO must verify whether to include security directly in its infrastructure in line with the distinguished architecture of its institution, risk exposure, and strategic security goals.

In all, the rapid adoption environment of Amnesty International by users and technical decision makers in the workplace seeks to obtain efficiency gains – are seduced by their personal use of consumer confrontation models such as ChatGPT, Microsoft Copilot, Claud Anthropor, Google Gemini, and others – even without clear guidelines of organizations from Organizations, are similar to Wish with with Thows Thow Thow The Thelding Agei. The adoption of the uninterrupted and unacceptable smartphones at the workplace during the “BYD” era in the early first decade of the twentieth and 2010 century.

However, in this case, the adoption curve of the Gen AI models between users is more slope and technology develops much faster, than many players, making it more than a safety mine field.

From interaction to the actual time: Why are the security matters integrated for obstetric artificial intelligence?

Traditional artificial intelligence safety tools that rely on external surveying and post -publication interventions make institutions weak at precise end points and threat surfaces when and where protection is more important.

Crowdstrike Integration in Falcon Cloud in Nvidia Universal LLM NIM converts this continuous dynamic defense directly in the life intelligence cycle from development to operation time.

Bernard also explained how AI-SPM from Falcon relieves proactive risk before publishing: “Falcon Cloud Security Ai-SPM gives security control and information technology teams early in this process-which helps to move for poor behavior, unauthorized models, and political violations before anything passes.

Falcon’s inclusion directly in the NVIDIA AI infrastructure leads to the automation of compliance with emerging regulations, such as the European Union Antilation law, making the integrity of the comprehensive model, tracking, auditing a fundamental part and automating from each publication instead of a manual and intense density task.

What is the integration of Crowdstrike with Nvidia for Cisos and Enterprise Gen Gen Ai Security

Trucific artificial intelligence quickly expands the surfaces of institutions attack, attracting ocean -based traditional security methods.

Threats to obstetric models including fast injection, data leakage, and typical poisoning all require a deeper vision, greater accuracy and control. The integration of Crowdstrike with the NVIDIA LLM infrastructure is noticeable to its architectural approach to address these security gaps.

For CISO, security leaders and Devops teams provide they serve, and include security controls directly in the artificial intelligence life cycle of concrete operational benefits including the following:

• The essential zero on a large scale: Automated publication of security policies eliminates manual effort and constantly enforcing zero confidence through each Amnesty International model.
• Continuous mitigation of weakness: Determining and neutralizing the risks before the operating time greatly reduces the windows of the attackers from opportunities.
• Constant operation time: The disclosure that is moved by the measurement is quickly determined in the actual time quickly and prevents threats such as immediate injection, typical poisoning, and nomination of unauthorized data.

Bernard emphasized the operational necessity to take a more integrated approach to artificial intelligence. “We focus on securing the models they adopt for institutions-especially those that are seized on sensitive or private data and help customers stay in the forefront because this technology becomes essential for how it works.”

Since artificial intelligence does not become mere discrimination but rather a basis for institutions infrastructure, the guaranteed safety is no longer optional. Crowdstrike and Nvidia not only adds protection; Reinfers to how to build artificial intelligence systems to withstand the already advanced tradecraft.