Daily Tech Digest - October 17, 2024

Digital addiction detox: Streamline tech to maximize impact, minimize risks

While digital addiction has been extensively studied at the individual level, organizational digital addiction is a relatively new area of concern. This addiction manifests as a tendency for the organization to throw technology mindlessly at any problem, often accumulating useless or misused technologies that generate ongoing costs without delivering proportional value. ... CIOs must simultaneously implement controls to prevent their organizations from reaching a tipping point where healthy exploration transforms into digital addiction. Striking this balance is delicate and requires careful management. Many innovative technology companies have found success by implementing “runways” for new products or technologies. These runways come with specific criteria for either “takeoff” or “takedown”. ... Unchecked technology adoption poses significant risks to organizations, often leading to vulnerabilities in their IT ecosystems. When companies rush to implement technologies without proper planning and safeguards, they lack the resilience to bounce back from adverse conditions because of insufficient redundancy and flexibility within systems, leaving organizations exposed to single points of failure.


Why are we still confused about cloud security?

A prevalent issue is publicly exposed storage, which often includes sensitive data due to excessive permissions, making it a prime target for ransomware attacks. Additionally, the improper use of access keys remains a significant threat, with a staggering 84% of organizations retaining unused highly privileged keys. Such security oversights have historically facilitated breaches, as evidenced by incidents like the MGM Resorts data breach in September 2023. ... Kubernetes environments present another layer of risk. The study notes that 78% of organizations have publicly accessible Kubernetes API servers, with significant portions allowing inbound internet access and unrestricted user control. This lax security posture exacerbates potential vulnerabilities. Addressing these vulnerabilities demands a comprehensive approach. Organizations should adopt a context-driven security ethos by integrating identity, vulnerability, misconfiguration, and data risk information. This unified strategy allows for precise risk assessment and prioritization. Managing Kubernetes access through adherence to Pod Security Standards and limiting privileged containers is essential, as is the regular audit of credentials and permissions to enforce the principle of least privilege.


The Architect’s Guide to Interoperability in the AI Data Stack

At the heart of an AI-driven world is data — lots of it. The choices you make today for storing, processing and analyzing data will directly affect your agility tomorrow. Architecting for interoperability means selecting tools that play nicely across environments, reducing reliance on any single vendor, and allowing your organization to shop for the best pricing or feature set at any given moment. ... Interoperability extends to query engines as well. Clickhouse, Dremio and Trino are great examples of tools that let you query data from multiple sources without needing to migrate it. These tools allow users to connect to a wide range of sources, from cloud data warehouses like Snowflake to traditional databases such as MySQL, PostgreSQL and Microsoft SQL Server. With modern query engines, you can run complex queries on data wherever it resides, helping avoid costly and time-consuming migrations. ... Architecting for interoperability is not just about avoiding vendor lock-in; it’s about building an AI data stack that’s resilient, flexible and cost-effective. By selecting tools that prioritize open standards, you ensure that your organization can evolve and adapt to new technologies without being constrained by legacy decisions. 


The role of compromised cyber-physical devices in modern cyberattacks

A cyber physical device is a device that connects the physical world and computer networks. Many people may associate the term “cyber physical device” with Supervisory Control and Data Acquisition (SCADA) systems and OT network segments, but there’s more to it. Devices that interconnect the physical world give attackers a unique perspective: they allow them to perform on-ground observation of events, to monitor and observe the impact of their attacks, and can even sometimes make an impact on the physical world ... Many devices are compromised for the simple purpose of creating points of presence at new locations, so attackers can bypass geofencing restrictions. These devices are often joined and used as a part of overlay networks. Many of these devices are not traditional routers but could be anything from temperature sensors to cameras. We have even seen compromised museum Android display boards in some countries. ... Realistically, I don’t believe there is a way to decrease number of compromised devices. We are moving towards networks where IoT devices will be one of the predominant types of connected devices, with things like a dish washer or fridge having an IP address. 


Security at the Edge Needs More Attention

CISOs should verify that the tools they acquire and use do what they claim to do, or they may be in for surprises. Meanwhile, data and IP are at risk because it’s so easy to sign up for and use third-party cloud services and SaaS that the average users may not associate their data usage with organizational risk. “Users submitting spreadsheet formula problems to online help forms may inadvertently be sharing corporate data. People running grammar checking tools on emails or documents may be doing the same,” says Roger Grimes, data-driven defense evangelist at security awareness training and simulated phishing platform KnowBe4 in an email interview. “It's far too easy for someone using an AI-enabled tool to not realize they are inadvertently leaking confidential information outside their organizational environment.”  ... It’s important for CISOs to have knowledge of and visibility into every asset in their company’s tech stack, though some CISOs see room for improvement. “You spend a lot of time and money on people, processes and technology to develop a layered security approach and defense in depth, and that doesn't work if you don't know you have something to defend there,” says Fowler.


CIOs must also serve as chief AI officers, according to Salesforce survey

CIOs are now in the business of manufacturing intelligence and work-autonomous work. CIOs are now responsible for creating a work environment where humans and AI agents can collaborate and co-create stakeholder value -- employees, customers, partners, and communities. CIOs must design, own, and deliver the roadmap to the autonomous enterprise, where autonomous work is maturing at Lightspeed. ... CIOs are under pressure to quickly learn about, and implement, effective AI solutions in their businesses. While more than three of five CIOs think stakeholder expectations for their AI expertise are unrealistic, only 9% believe their peers are more knowledgeable. CIOs are also partnering with analyst firms (Gartner, Forrester, IDC, etc.) and technology vendors to learn more about AI. ... Sixty-one percent of CIOs feel they're expected to know more about AI than they do, and their peers at other companies are their top sources of information. CIOs must become better AI storytellers. In 1994, Steve Jobs said: "The most powerful person in the world is the storyteller. The storyteller sets the vision, values, and agenda of an entire generation that is to come." There is no better time than now for CIOs to lead the business transformation towards becoming AI-led companies.


Policing and facial recognition: What’s stopping them?

The question contains two “ifs” and a presumption; all are carrying a lot of weight. The first “if” is the legal basis for using FRT. Do the police have the power to use it? In England and Wales the police certainly have statutory powers to take and retain images of people, along with common law powers to obtain and store information about the citizen’s behavior in public. The government’s own Surveillance Camera Code of Practice (currently on policy’s death row) provides guidance to chief officers on how to do this and on operating overt surveillance systems in public places generally. The Court of Appeal found a “sufficient legal framework” covered police use of FRT, one that was capable of supporting its lawful deployment. ... The second “if” relates to the technology i.e. “if FRT works, what’s stopping the police from using it?” Since a shaky introduction around 2015 when it didn’t work as hoped (or required) police facial recognition technology has come on significantly. The accuracy of the technology is much better but is it accurate to say it now “works”? Each technology partner and purchasing police force must answer that for themselves – as for any other operational capability. That’s accountability. 


How AI is becoming a powerful tool for offensive cybersecurity practitioners

What makes offensive security all the more important is that it addresses a potential blind spot for developers. “As builders of software, we tend to think about using whatever we’ve developed in the ways that it’s intended to be used,” says Caroline Wong, chief strategy officer at Cobalt Labs, a penetration testing company. In other words, Wong says, there can be a bias towards overemphasizing the good ways in which software can be used, while overlooking misuse and abuse cases or disregarding potentially harmful uses. “One of the best ways to identify where and how an organization or a piece of software might be susceptible to attack is by taking on the perspective of a malicious person: the attacker’s mindset,” Wong says. ... In addition to addressing manpower issues, AI can assist practitioners in scaling up their operations. “AI’s ability to process vast datasets and simulate large-scale attacks without human intervention allows for testing more frequently and on a broader scale,” says Augusto Barros, a cyber evangelist at Securonix, a security analytics and operations management platform provider. “In large or complex environments, human operators would struggle to perform consistent and exhaustive tests across all systems,” Barros says. 


While Cyberattacks Are Inevitable, Resilience Is Vital

Cybersecurity is all about understanding risk and applying the basic controls and sprinkling in new technologies to keep the bad guys out and keeping the system up and running by eliminating as much unplanned downtime as possible. “Cybersecurity is a risk game—as long as computers are required to deliver critical products and services, they will have some vulnerability to an attack,” Carrigan said. “Risk is a simple equation: Risk = Likelihood x Consequence. Most of our investments have been in reducing the ‘likelihood’ side of the equation. The future of OT cybersecurity will be in reducing the consequences of cyberattacks—specifically, how to minimize the impact of infiltration and restore operations within an acceptable period.” Manufacturers must understand their risk appetite and know what and where their organization’s crown jewels are and how to protect them. “Applying the same security practices to all OT assets is not practical—some are more important than others, even within the same company and the same OT network,” Carrigan said. Remaining resilient to a cyber incident—any kind of incident—means manufacturers must apply the basics, sprinkle in some new technologies and plan, test, revise and then start that process all over again. 


AI-Powered DevOps: Best Practices for Business Adoption

In security, AI tools are proving highly effective at proactively identifying and addressing vulnerabilities, boosting threat detection capabilities, and automating responses to emerging risks. Nonetheless, significant potential for AI remains in phases such as release management, deployment, platform engineering, and planning. These stages, which are crucial for ensuring software stability and scalability, could greatly benefit from AI's predictive abilities, resource optimization, and the streamlining of operational and maintenance processes. ... While generative AI and AI copilots have been instrumental in driving adoption of this technology, there remains a major shortage of AI expertise within DevOps. This gap is significant, especially given that humans remain deeply involved in the process, with over two-thirds of our respondents indicating they manually review AI-generated outputs at least half the time. To address these challenges, organizations should devise specialized training courses to properly equip their DevOps teams with the skills to leverage AI tools. Whether through industry-recognized courses or internal programs, encouraging certification can enhance technical expertise significantly.



Quote for the day:

"All progress takes place outside the comfort zone." -- Michael John Bobak

No comments:

Post a Comment