Daily Tech Digest - March 02, 2025

---

Quote for the day:

"The ability to summon positive emotions during periods of intense stress lies at the heart of effective leadership." -- Jim Loehr

---

Weak cyber defenses are exposing critical infrastructure — how enterprises can proactively thwart cunning attackers to protect us all

Weak cybersecurity isn’t merely a corporate issue — it’s a national security risk. The 2021 Colonial Pipeline attack disrupted energy supplies and exposed vulnerabilities in critical industries. Rising geopolitical tensions, especially with China, amplify these risks. Recent breaches attributed to state-sponsored actors have exploited outdated telecommunications equipment and other legacy systems, revealing how complacency in updating technology can put national security in danger. For instance, last year’s hack of U.S. and international telecommunications companies exposed phone lines used by top officials and compromised data from systems for surveillance requests, threatening national security. Weak cybersecurity at these companies risks long-term costs, allowing state-sponsored actors to access sensitive information, influence political decisions and disrupt intelligence efforts. ... No company can face today’s cyber threats on its own. Collaboration between private businesses and government agencies is more than helpful — it’s imperative. Sharing threat intelligence in real-time allows organizations to respond faster and stay ahead of emerging risks. Public-private partnerships can also level the playing field by offering smaller companies access to resources like funding and advanced security tools they might not otherwise afford.

Evaluating the CISO

Delegation skills are an essential component that should be evaluated separately in this area. Effective delegation is essential to prevent becoming a bottleneck, as micromanagement is unsuitable for the CISO role. Delegating complex tasks not only lightens your load but also helps foster the team’s overall competence. Without strong delegation skills, CISOs cannot rate themselves highly in their relationship with the internal security team. ... A CISO is hired to lead, manage, and support specific projects or programs such as migrating to a cloud or hybrid infrastructure, implementing zero-trust principles, launching security awareness initiatives, or assessing risks and creating a roadmap for post-quantum cryptography implementation. The success of these initiatives ultimately falls under the CISO’s responsibility. To execute these programs effectively, the CISO relies heavily on its team and internal organizational peers. As such, building strong relationships with both is essential for successfully delivering projects. ... A CISO must have responsibility for the information security budget, which includes funding for the team, tools, and services. Without direct control over the budget, it becomes challenging to rate the relationship with management highly, as budget ownership is a critical aspect of the CISO’s role.

Unraveling Large Language Model Hallucinations

You might have seen model hallucinations. They are the instances where LLMs generate incorrect, misleading, or entirely fabricated information that appears plausible. These hallucinations happen because LLMs do not “know” facts in the way humans do; instead, they predict words based on patterns in their training data. ... Supervised Fine-Tuning makes the model capable. However, even a well-trained model can generate misleading, biased, or unhelpful responses. Therefore, Reinforcement Learning with Human Feedback is required to align it with human expectations. We start with the assistant model, trained by SFT. For a given prompt we generate multiple model outputs. Human labelers rank or score multiple model outputs based on quality, safety, and alignment with human preferences. We use these data to train a whole separate neural network that we call a reward model. The reward model imitates human scores. It is a simulator of human preferences. It is a completely separate neural network, probably with a transformer architecture, but it is not a language model in the sense that it generates diverse language. It’s just a scoring model.

How to Communicate the Business Value of Master Data Management

In an ideal scenario, MDM is integral to a broader D&A strategy, highlighting how D&A supports the organization's strategic goals. The strategy aligns with these goals, prioritizes the business outcomes it will support, and details what is needed to achieve them. Therefore, leaders must first understand and prioritize the explicit business outcomes that MDM will support before creating an MDM strategy. In other words, "improving decision-making" is not good enough. "Increase customer service levels by 5% by end of December 2025" is the level of detail required. D&A leaders may recognize that master data is causing a problem or limiting an opportunity, which is where they would rely on an MDM. If this is the case, those D&A leaders should consider questions that help identify the problem, KPIs, and key stakeholders in these cases. These questions help identify potential business outcomes that MDM could support. Figure 1 provides a worksheet to build this initial picture and facilitate stakeholder discussions. The worksheet maps high-level goals onto a run-grow-transform framework, which could also be represented by three columns for the primary business value drivers: risk, revenue, and cost.

4 ways to get your business ready for the agentic AI revolution

Agents could be used eventually, but only once a partnership approach identifies the right opportunities. "Agents are becoming a big part of how generative AI and machine learning are used in business today. The way agents will be used in travel will be fascinating to watch. I think this technology will certainly be a part of the mix," he said. "The process for Hyatt will be to find the right technologies -- and we'll do that in close partnership with our business leaders and the technology teams that run the applications. We'll then provide the AI services to drive those transitions for the business." ... Keith Woolley, chief digital and information officer at the University of Bristol, is another digital leader who sees the potential benefits of agents. However, he said these advantages will become manifest over the longer term. "We are looking at agentic AI, but we're not implementing it yet," he said. "We sit as a management team and ask questions like, 'Should we do our admissions process using agentic AI? What would be the advantage?'" Woolley told ZDNET he could envision a situation in which AI and automation help assess and inform candidates worldwide about the status of their applications.

Cloud Giants Collaborate on New Kubernetes Resource Management Tool

The core innovation of kro is the introduction of the ResourceGraphDefinition custom resource. kro encapsulates a Kubernetes deployment and its dependencies into a single API, enabling custom end-user interfaces that expose only the parameters applicable to a non-platform engineer. This masking hides the complexity of API endpoints for Kubernetes and cloud providers that are not useful in a deployment context. ... Kro works seamlessly with the existing cloud provider Kubernetes extensions that are available to manage cloud resources from Kubernetes. These are AWS Controllers for Kubernetes (ACK), Google's Config Connector (KCC), and Azure Service Operator (ASO). kro enables standardised, reusable service templates that promote consistency across different projects and environments, with the benefit of being entirely Kubernetes-native. It is still in the early stages of development. "As an early-stage project, kro is not yet ready for production use, but we still encourage you to test it out in your own Kubernetes development environments," the post states. ... Most significantly for the Crossplane community, Farcic questioned kro's purpose given its functional overlap with existing tools. "kro is serving more or less the same function as other tools created a while ago without any compelling improvement," he observed. 

Why a different approach to AIOps is needed for SD-WAN

AIOps tools enhance efficiency by seamlessly integrating with IT management tools, enabling proactive issue identification and streamlining IT management processes. But more than that, they optimize an organization’s network by improving the performance, efficiency, and dependability of its network resources to ensure optimal user experience. Regarding infrastructure, many organizations now rely on SD-WAN – software-defined wide area network – to manage and optimize data traffic across different types of networks efficiently. SD-WAN is an effective way to connect the organization and provide users with application access. It helps businesses improve their network performance, cut costs, and be more flexible by easily connecting to various network types. ... AIOps tools use the information extracted from SD-WAN systems and autonomously resolve issues without human intervention. In other words, AIOps tools utilize predictive analytics to forecast future events or outcomes related to network operations. This makes the whole system run smoother and more reliably, while machine learning algorithms can use this historical data to make predictions and proactively improve the performance of critical applications.

AI-Driven Threat Detection and the Need for Precision

AI algorithms, particularly those based on machine learning, excel at sifting through massive datasets and identifying patterns that would be nearly impossible for us mere humans to spot. An AI system might analyze network traffic patterns to identify unusual data flows that could indicate a data exfiltration attempt. Alternatively, it could scan email attachments for malicious code that traditional antivirus software might miss. Ultimately, AI feeds on context and content. The effectiveness of these systems in protecting your security posture(link is external) is inextricably linked to the quality of the data they are trained on and the precision of their algorithms. ... Finally, AI-driven threat detection may not eradicate human expertise. Skilled security professionals should still oversee AI systems and make informed decisions based on their own contextual expertise and experience. Human oversight validates the AI's findings, and threat detection algorithms may not be able to totally replace the critical thinking and intuition of human analysts. There may come a time when human professionals exist in AI's shadow. Yet, at this time, combining the power of AI with human knowledge and a commitment to continuous learning can form the building blocks for a sophisticated defense program. 

From Ambiguity to Accountability: Analyzing Recommender System Audits under the DSA

In these early years of the DSA, a range of stakeholders – online platforms, civil society, the European Commission (EC), and national Digital Service Coordinators (DSCs) – must experiment, identify good practices, and share lessons learned. Such iteration is important to ensure an adaptive DSA regime that spurs innovation and responds to shifting technologies, risks, and mitigation strategies. The need for iteration and flexibility, however, should not mean the audits fail to deliver on their potential as vehicles for transparency and accountability. The first round of independent audits of recommender systems reveals clear areas for immediate improvement. Because the core definitions and methodologies were developed independently by platforms and auditors, significant inconsistencies exist in both risk assessment and audit processes. ... The DSA requires the main parameters of recommender systems to be spelled out in plain and intelligible language. What does this concretely mean in the recommender system context? Is it free of “acronyms or complex/technical terminology” (Pinterest), “straightforward vocabulary and easy to perceive, understand, or interpret” (Snap), or “written for a general audience with varying technical skill levels, inclusive of all users” (TikTok)? There's a subtle difference in expectations associated with each framing. These terms don’t need to be defined in a vacuum.

Cybersecurity in retail: What does the future hold?

In the coming year, cybersecurity experts predict attackers will increasingly target Generative AI models used by retailers, creating significant potential for operational disruptions and data breaches. These AI systems, now critical to retail operations, are vulnerable to sophisticated attacks that could compromise customer service efficiency and expose critical business vulnerabilities. The core risk lies in the sophisticated ways attackers can exploit AI’s complex decision-making processes, turning what was once a technological advantage into a potential security liability. Retailers must recognise that their AI systems are not just technological tools, but potential entry points for cybercriminal activities. ... The complexity and distribution of digital ecosystems make them prime targets during high-demand periods. For example, as we have seen in the past, cyberattacks that hit supply chains can cause major delays and financial loss. These incidents underscore the vulnerabilities in supply chains during peak times of the year​. In 2025, expect a rise in supply chain attacks during the holiday season, targeting ecommerce platforms and logistics providers, which could disrupt product availability and shipping.