The Race to Secure Artificial Intelligence

23 5 minutes read

Over the past few years, the world has been fascinated by the creative and intellectual power of artificial intelligence (AI). We’ve watched her create art, write code, and discover new medicines. And now, as of October 2025, we are handing her the keys to the kingdom. AI is no longer just a cool tool; It is the operating brain of our energy grids, financial markets and logistics networks. We’re building a digital god in a box, but we’ve barely begun to ask the most important question of all: How do we protect it from corruption, theft, or turning against us? The field of AI cybersecurity is not just another IT sub-discipline; It is the most important security challenge of the 21st century.

New Attack Surface: Mind Hacking

Securing artificial intelligence is fundamentally different from securing a traditional computer network. A hacker doesn’t need to break through a firewall if he can manipulate the “mind” of the AI itself. The attack vectors are subtle, insidious and entirely new. Primary threats include:

Data poisoning: This is the most insidious attack. The adversary subtly inserts biased or malicious data into the massive data sets used to train the AI. The result is a compromised model that appears to work normally but has a hidden, exploitable flaw. Imagine that AI trained to detect financial fraud is secretly taught that transactions from a particular criminal enterprise are always legitimate.
Extract form: This is the new industrial espionage. Adversaries can use complex queries to “steal” a multi-billion-dollar proprietary AI model by reverse-engineering its behavior, allowing them to replicate it for their own purposes.
Instantaneous injections and hostile attacks: This is the most common threat, as users craft clever prompts to trick live AI into bypassing its security protocols, revealing sensitive information, or executing malicious commands. A study by the Artificial Intelligence Security Research Consortium showed that this is already a rampant problem.
Supply chain attacks: AI models are not built from scratch; They are built using open source libraries and pre-trained components. A vulnerability introduced into a popular machine learning library could create a backdoor into thousands of artificial intelligence systems.

Human approach versus AI approach

Two main philosophies have emerged to address this unprecedented challenge.

The first is the human-driven “castle” model. This is the traditional approach to cybersecurity, adapted for artificial intelligence. It involves strict human oversight, with teams of experts conducting penetration tests, reviewing training data for signs of toxicity, and establishing strict ethical and operational guardrails. “Red teams” of human hackers are employed to find vulnerabilities and patch them before they are exploited. This approach is intentional, auditable, and grounded in human ethics. But its main weakness is speed. A human team cannot simply review a trillion-point data set in real-time or confront an AI-driven attack that evolves in milliseconds.

The second is the “immune system” model driven by artificial intelligence. This approach assumes that the only thing that can effectively defend an AI is another AI. This “sentinel AI” will act as a biological immune system, constantly monitoring the underlying AI for anomalous behavior, detecting subtle signs of data poisoning, and identifying and neutralizing hostile attacks in real time. This model provides the speed and scale needed to confront modern threats. Her big and terrifying weakness is “Who’s watching the Watchers?” problem. If the Guardian’s AI itself is compromised, or if its definition of “malicious” behavior becomes skewed, it could become an even greater threat.

Verdict: Coexistence of human and artificial intelligence

The debate over whether people or AI should lead these efforts is a false choice. The only viable path forward is a deep symbiotic partnership. We must build a system in which AI is the front-line soldier and humans are the strategic commander.

Sentinel AI must handle high-volume defense in real-time: scanning trillions of data points, flagging suspicious queries, and patching low-level vulnerabilities at machine speed. Human experts, in turn, must develop the strategy. It sets moral red lines, designs the security architecture, and, most importantly, serves as the ultimate authority to make critical decisions. If the guardian AI detects a major system-wide attack, it should not act unilaterally; It must isolate the threat and alert a human operator who makes the final call. As the federal Cybersecurity and Infrastructure Security Agency (CISA) explains, this “human-in-the-loop” model is essential to maintaining control.

National strategy for artificial intelligence security

This is not a problem that companies can solve alone; It is a matter of national security. The state’s strategy must be multifaceted and decisive.

Establishment of a National Artificial Intelligence Security Center (NAISC): A public-private partnership, similar to DARPA to defend AI, to fund research, develop best practices, and serve as a threat intelligence center.
Third-party audit authorization: Just as the SEC requires financial audits, the government should require that all companies deploying “AI for critical infrastructure” (e.g. for energy or finance) undergo regular, independent security audits by accredited companies.
Investing in talent: We must fund university programs and create professional degrees to develop a new category of expert: the AI security specialist, a hybrid expert in both machine learning and cybersecurity.
Promoting international standards: AI threats are global. The United States should lead the charge in establishing international treaties and rules for the safe and ethical development of artificial intelligence, similar to the nuclear non-proliferation treaties.

Securing the hybrid AI enterprise: Lenovo’s strategic framework

Lenovo is aggressively positioning itself as a trusted architect of enterprise AI by leveraging its deep heritage and focus on security and end-to-end execution, a strategy that currently outperforms competitors like Dell. Their approach, Lenovo Hybrid AI Advantage, is a complete framework designed to ensure customers not only deploy AI, but also achieve measurable ROI and ensure security. Key to this is addressing the human element through new AI adoption and change management services, recognizing that improving workforce skills is essential to effectively scaling AI.

Furthermore, Lenovo addresses the massive computational demands of AI with physical flexibility. Its leadership in integrating liquid cooling into data center infrastructure (New 6th Generation Neptune® Liquid Cooling for AI Tasks – Lenovo) is a significant competitive advantage, enabling denser, more energy-efficient AI factories that are vital to running powerful large language models (LLMs). By combining these trusted infrastructure solutions with robust security and verified vertical AI solutions – from workplace safety to retail analytics – Lenovo positions itself as a partner that provides not just hardware, but the complete and secure ecosystem needed for a successful AI transformation. This combination of IBM’s legacy enterprise focus and cutting-edge thermal management makes Lenovo a uniquely strong choice for securing the future of complex hybrid AI.

wrap

The power of artificial intelligence is growing at an exponential rate, but our strategies for securing it are seriously lagging behind. The threats are no longer theoretical. The solution is not a choice between humans and AI, but rather a combination of human strategic oversight and real-time defense powered by AI. For a country like the United States, developing a comprehensive national strategy to secure its AI infrastructure is not optional. It is the essential requirement to ensure that the most powerful technology ever created remains a tool for progress, not a weapon for catastrophic failure, and Lenovo may be the most qualified supplier to assist in this effort.

As President and Principal Analyst of The Enderle Group, Rob provides regional and global companies with guidance on how to create a trusted dialogue with the marketplace, target customer needs, create new business opportunities, anticipate technological changes, select vendors and products, and practice zero-dollar marketing. Rob has worked for over 20 years at and with companies such as Microsoft, HP, IBM, Dell, Toshiba, Gateway, Sony, USAA, Texas Instruments, AMD, Intel, Credit Suisse First Boston, ROLM, and Siemens.