An Update on AI – Rapid Rise and Security Risks

In a very short time, AI tools have become ubiquitous and are radically changing the way we live and work. From daily tasks such as searching for information or summarizing reports to complex professional assignments, AI seems to be everywhere.

That’s why, last week, the place to be was San Jose, California, for Nvidia’s GTC AI Conference—either in person (even better) or virtually (as I did). GTC stands for GPU (Graphics Processing Unit) Technology Conference, originally focused on the gaming industry until researchers began harnessing the power of GPUs for machine learning and AI.

This year, Nvidia CEO Jensen Huang delivered the keynote in a fully packed sports stadium, prompting many to call the event the “Super Bowl of AI.” In an impressive two-hour presentation, Huang traced AI’s evolution over the past decade—from perception and computer vision to generative AI and now “Agentic AI,” which incorporates reasoning and contextual understanding (see one explanation of Agentic AI here). He argued that this progression is fundamentally transforming computing and announced upcoming generations of significantly more powerful GPUs.

One of the major challenges to AI’s continued progress, he noted, is data availability. To address this, Nvidia is investing heavily in synthetic data generation, allowing models to train on simulated experiences, reducing reliance on human-generated data and speeding up development.

The conference attracted over 25,000 attendees and the atmosphere was overwhelmingly optimistic. Topics included autonomous vehicles and robotics, all underscoring the accelerating momentum of AI innovation.

Yet one critical topic remained mostly in the background: the potential for rogue or malicious use of AI.

Given recent events in the Middle East—especially since October 7, 2023—it’s easy to imagine how AI could be weaponized or used to carry out destructive goals. The concern is not theoretical. The U.S. recently issued a regulation titled the Framework for Artificial Intelligence Diffusion, aimed at preventing cutting-edge technologies from reaching adversarial actors. As the document (released January 2025) notes, “…as the capabilities of these models continue to improve, they will enable malicious actors to engage in activities that pose profound risks to U.S. national security and foreign policy objectives.”

Still, it remains uncertain whether such regulations will effectively prevent embargoed nations—like China, Russia, or Iran—from acquiring high-performance AI capabilities. One recent report on the U.S.–China AI race concludes that while China may have millions of chips in the next couple of years, they remain far inferior to Nvidia’s GPUs for advanced training tasks (see here).

However, there is a major difference between a “profound risk to the U.S.” and the threat posed by AI in the hands of terror groups. Sadly, it’s plausible that some groups may already be leveraging AI—much like they eventually adopted drone technology as a staple of asymmetric warfare—but this time, the adoption may happen much faster (see this study on the risks of Generative AI misuse).

Israel, meanwhile, is still working toward a comprehensive national AI strategy to ensure its global competitiveness. The Israel Innovation Authority is spearheading a National AI Program (see here), but more is needed. As recent commentary has noted (see interview), Israel must act swiftly and decisively to stay at the forefront—not only for its economic future, but also to address growing national security challenges.