Daily Tech Digest - August 26, 2025


Quote for the day:

“When we give ourselves permission to fail, we, at the same time, give ourselves permission to excel.” -- Eloise Ristad


6 tips for consolidating your vendor portfolio without killing operations

Behind every sprawling vendor relationship is a series of small extensions that compound over time, creating complex entanglements. To improve flexibility when reviewing partners, Dovico is wary of vendor entanglements that complicate the ability to retire suppliers. Her aim is to clearly define the service required and the vendor’s capabilities. “You’ve got to be conscious of not muddying how you feel about the performance of one vendor, or your relationship with them. You need to have some competitive tension and align core competencies with your problem space,” she says. Klein prefers to adopt a cross-functional approach with finance and engineering input to identify redundancies and sprawl. Engineers with industry knowledge cross-reference vendor services, while IT checks against industry benchmarks, such as Gartner’s Magic Quadrant, to identify vendors providing similar services or tools. ... Vendor sprawl also lurks in the blind spot of cloud-based services that can be adopted without IT oversight, fueling shadow purchasing habits. “With the proliferation of SaaS and cloud models, departments can now make a few phone calls or sign up online to get applications installed or services procured,” says Klein. This shadow IT ecosystem increases security risks and vendor entanglement, undermining consolidation efforts. This needs to be tackled through changes to IT governance.


Should I stay or should I go? Rethinking IT support contracts before auto-renewal bites

Contract inertia, which is the tendency to stick with what you know, even when it may no longer be the best option, is a common phenomenon in business technology. There are several reasons for it, such as familiarity with an existing provider, fear of disruption, the administrative effort involved in reviewing and comparing alternatives, and sometimes just a simple lack of awareness that the renewal date is approaching. The problem is that inertia can quietly erode value. As organisations grow, shift priorities or adopt new technologies, the IT support they once chose may no longer be fit for purpose. ... A proactive approach begins with accountability. IT leaders need to know what their current provider delivers and how they are being used by the company. Are remote software tools performing as expected? Are updates, patches and monitoring processes being applied consistently across all platforms? Are issues being resolved efficiently by our internal IT team, or are inefficiencies building up? Is this the correct set-up and structure for our business, or could we be making better use of existing internal capacity, by leveraging better remote management tools? Gathering this information allows organisations to have an honest conversation with their provider (and themselves) about whether the contract still aligns with their objectives.


AI Data Security: Core Concepts, Risks, and Proven Practices

Although AI makes and fortifies a lot of our modern defenses, once you bring AI into the mix, the risks evolve too. Data security (and cybersecurity in general) has always worked like that. The security team gets a new tool, and eventually, the bad guys get one too. It’s a constant game of catch-up, and AI doesn’t change that dynamic. ... One of the simplest ways to strengthen AI data security is to control who can access what, early and tightly. That means setting clear roles, strong authentication, and removing access that people don’t need. No shared passwords. No default admin accounts. No “just for testing” tokens sitting around with full privileges. ... What your model learns is only as good (and safe) as the data you feed it. If the training pipeline isn’t secure, everything downstream is at risk. That includes the model’s behavior, accuracy, and resilience against manipulation. Always vet your data sources. Don’t rely on third-party datasets without checking them for quality, bias, or signs of tampering. ... A core principle of data protection, baked into laws like GDPR, is data minimization: only collect what you need, and only keep it for as long as you actually need it. In real terms, that means cutting down on excess data that serves no clear purpose. Put real policies in place. Schedule regular reviews. Archive or delete datasets that are no longer relevant. 


Morgan Stanley Open Sources CALM: The Architecture as Code Solution Transforming Enterprise DevOps

CALM enables software architects to define, validate, and visualize system architectures in a standardized, machine-readable format, bridging the gap between architectural intent and implementation. Built on a JSON Meta Schema, CALM transforms architectural designs into executable specifications that both humans and machines can understand. ... The framework structures architecture into three primary components: nodes, relationships, and metadata. This modular approach allows architects to model everything from high-level system overviews to detailed microservices architectures. ... CALM’s true power emerges in its seamless integration with modern DevOps workflows. The framework treats architectural definitions like any other code asset, version-controlled, testable, and automatable. Teams can validate architectural compliance in their CI/CD pipelines, catching design issues before they reach production. The CALM CLI provides immediate feedback on architectural decisions, enabling real-time validation during development. This shifts compliance left in the development lifecycle, transforming potential deployment roadblocks into preventable design issues. Key benefits for DevOps teams include machine-readable architecture definitions that eliminate manual interpretation errors, version control for architectural changes that provides clear change history, and real-time feedback on compliance violations that prevent downstream issues.


Shadow AI is surging — getting AI adoption right is your best defense

Despite the clarity of this progression, many organizations struggle to begin. One of the most common reasons is poor platform selection. Either no tool is made available, or the wrong class of tool is introduced. Sometimes what is offered is too narrow, designed for one function or team. Sometimes it is too technical, requiring configuration or training that most users aren’t prepared for. In other cases, the tool is so heavily restricted that users cannot complete meaningful work. Any of these mistakes can derail adoption. A tool that is not trusted or useful will not be used. And without usage, there is no feedback, value, or justification for scale. ... The best entry point is a general-purpose AI assistant designed for enterprise use. It must be simple to access, require no setup, and provide immediate value across a range of roles. It must also meet enterprise requirements for data security, identity management, policy enforcement, and model transparency. This is not a niche solution. It is a foundation layer. It should allow employees to experiment, complete tasks, and build fluency in a way that is observable, governable, and safe. Several platforms meet these needs. ChatGPT Enterprise provides a secure, hosted version of GPT-5 with zero data retention, administrative oversight, and SSO integration. It is simple to deploy and easy to use. =


AI and the impact on our skills – the Precautionary Principle must apply

There is much public comment about AI replacing jobs or specific tasks within roles, and this is often cited as a source of productivity improvement. Often we hear about how junior legal professionals can be easily replaced since much of their work is related to the production of standard contracts and other documents, and these tasks can be performed by LLMs. We hear much of the same narrative from the accounting and consulting worlds. ... The greatest learning experiences come from making mistakes. Problem-solving skills come from experience. Intuition is a skill that is developed from repeatedly working in real-world environments. AI systems do make mistakes and these can be caught and corrected by a human, but it is not the same as the human making the mistake. Correcting the mistakes made by AI systems is in itself a skill, but a different one. ... In a rapidly evolving world in which AI has the potential to play a major role, it is appropriate that we apply the Precautionary Principle in determining how to automate with AI. The scientific evidence of the impact of AI-enabled automation is still incomplete, but more is being learned every day. However, skill loss is a serious, and possibly irreversible, risk. The integrity of education systems, the reputations of organisations and individuals, and our own ability to trust in complex decision-making processes, are at stake.


Ransomware-Resilient Storage: The New Frontline Defense in a High-Stakes Cyber Battle

The cornerstone of ransomware resilience is immutability: data written to storage cannot be altered or deleted ever. This write-once-read-many capability means backup snapshots or data blobs are locked for prescribed retention periods, impervious to tampering even by attackers or system administrators with elevated privileges. Hardware and software enforce this immutability by preventing any writes or deletes on designated volumes, snapshots, or objects once committed, creating a "logical air gap" of protection without the need for physical media isolation. ... Moving deeper, efforts are underway to harden storage hardware directly. Technologies such as FlashGuard, explored experimentally by IBM and Intel collaborations, embed rollback capabilities within SSD controllers. By preserving prior versions of data pages on-device, FlashGuard can quickly revert files corrupted or encrypted by ransomware without network or host dependency. ... Though not widespread in production, these capabilities signal a future where storage devices autonomously resist ransomware impact, a powerful complement to immutable snapshotting. While these cutting-edge hardware-level protections offer rapid recovery and autonomous resilience, organizations also consider complementary isolation strategies like air-gapping to create robust multi-layered defense boundaries against ransomware threats.


How an Internal AI Governance Council Drives Responsible Innovation

The efficacy of AI governance hinges on the council’s composition and operational approach. An optimal governance council typically includes cross-functional representation from executive leadership, IT, compliance and legal teams, human resources, product management, and frontline employees. This diversified representation ensures comprehensive coverage of ethical considerations, compliance requirements, and operational realities. Initial steps in operationalizing a council involve creating strong AI usage policies, establishing approved tools, and developing clear monitoring and validation protocols. ... While initial governance frameworks often focus on strict risk management and regulatory compliance, the long-term goal shifts toward empowerment and innovation. Mature governance practices balance caution with enablement, providing organizations with a dynamic, iterative approach to AI implementation. This involves reassessing and adapting governance strategies, aligning them with evolving technologies, organizational objectives, and regulatory expectations. AI’s non-deterministic, probabilistic nature, particularly generative models, necessitates a continuous human oversight component. Effective governance strategies embed this human-in-the-loop approach, ensuring AI enhances decision-making without fully automating critical processes.


The energy sector has no time to wait for the next cyberattack

Recent findings have raised concerns about solar infrastructure. Some Chinese-made solar inverters were found to have built-in communication equipment that isn’t fully explained. In theory, these devices could be triggered remotely to shut down inverters, potentially causing widespread power disruptions. The discovery has raised fears that covert malware may have been installed in critical energy infrastructure across the U.S. and Europe, which could enable remote attacks during conflicts. ... Many OT systems were built decades ago and weren’t designed with cyber threats in mind. They often lack updates, patches, and support, and older software and hardware don’t always work with new security solutions. Upgrading them without disrupting operations is a complex task. OT systems used to be kept separate from the Internet to prevent remote attacks. Now, the push for real-time data, remote monitoring, and automation has connected these systems to IT networks. That makes operations more efficient, but it also gives cybercriminals new ways to exploit weaknesses that were once isolated. Energy companies are cautious about overhauling old systems because it’s expensive and can interrupt service. But keeping legacy systems in play creates security gaps, especially when connected to networks or IoT devices. Protecting these systems while moving to newer, more secure tech takes planning, investment, and IT-OT collaboration.


Agentic AI Browser an Easy Mark for Online Scammers

In an Wednesday blog post, researchers from Guardio wrote that Comet - one of the first AI browsers to reach consumers - clicked through fake storefronts, submitted sensitive data to phishing sites and failed to recognize malicious prompts designed to hijack its behavior. The Tel Aviv-based security firm calls the problem "scamlexity," a messy intersection of human-like automation and old-fashioned social engineering creates "a new, invisible scam surface" that scales to millions of potential victims at once. In a clash between the sophistication of generative models built into browsers and the simplicity of phishing tricks that have trapped users for decades, "even the oldest tricks in the scammer's playbook become more dangerous in the hands of AI browsing." One of the headline features of AI browsers is one-click shopping. Researchers spun up a fake "Walmart" storefront complete with polished design, realistic listings and a seamless checkout flow. ... Rather than fooling a user into downloading malicious code to putatively fix a computer problem - as in ClickFix - a PromptFix attack is a malicious instruction was hidden inside what looks like a CAPTCHA. The AI treated the bogus challenge as routine, obeyed the hidden command and continued execution. AI agents are expected to ingest unstructured logs, alerts or even attacker-generated content during incident response.

No comments:

Post a Comment