Daily Tech Digest - January 23, 2025


Quote for the day:

"Great leaders go forward without stopping, remain firm without tiring and remain enthusiastic while growing" -- Reed Markham


Cyber Insights 2025: APIs – The Threat Continues

APIs are easily written, often with low-code / no-code tools. They are often considered by the developer as unimportant in comparison to the apps they connect, and probably protected by the tools that protect the apps. Bad call. “API attacks will increase in 2025 due to this over-reliance on existing application security and API management tools, but also due to organizations dragging their heels when it comes to protecting APIs,” says James Sherlow, systems engineering director of EMEA at Cequence Security. “While there was plenty of motivation to roll out APIs to stand up new services and support revenue streams, the same incentives are not there when it comes to protecting them.” Meanwhile, attackers are becoming increasingly sophisticated in their attacks. “In contrast, threat actors are not resting on their laurels,” he continued. “It’s now not uncommon for them to use multi-faceted attacks that seek to evade detection and then dodge and feint when the attack is blocked, all the time waiting until the last minute to target their end goal.” In short, he says, “It’s not until the business is breached that it wakes up to the fact that API protection and application protection are not one and the same thing. Web Application Firewalls, Content Delivery Networks, and API Gateways do not adequately protect APIs.”


Box-Checking or Behavior-Changing? Training That Matters

The pressure to meet these requirements is intense, and when a company finds an “acceptable” solution, they too often just check the box knowing they are compliant and stick with that solution in perpetuity - whether it creates a more secure workplace and behavioral change or not. Training programs designed purely to meet regulations are rarely effective. These initiatives tend to rely on generic content that employees skim through and forget. Organizations may meet the legal standard, but they fail to address the root causes of risky behavior. ... To improve outcomes, training programs must connect with people on a more practical level. Tailoring the content to fit specific roles within the organization is one way to do this. The threats a finance team faces, for example, are different from those encountered by IT professionals, so their training should reflect those differences. When employees see the relevance of the material, they are more likely to engage with it. Professionals in security awareness roles can distinguish themselves by designing programs that meet these needs. Equally important is embracing the concept of continuous learning. Annual training sessions often fail to stick. Smaller, ongoing lessons delivered throughout the year help employees retain information and incorporate it into their daily routines. 


OpenAI opposes data deletion demand in India citing US legal constraints

OpenAI has informed the Delhi High Court that any directive requiring it to delete training data used for ChatGPT would conflict with its legal obligations under US law. The statement came in response to a copyright lawsuit filed by the Reuters-backed Indian news agency ANI, marking a pivotal development in one of the first major AI-related legal battles in India. ... This case mirrors global legal trends, as OpenAI faces similar lawsuits in the United States and beyond, including from major organizations like The New York Times. OpenAI maintains its position that it adheres to the “fair use” doctrine, leveraging publicly available data to train its AI systems without infringing intellectual property laws. In the case of Raw Story Media v. OpenAI, heard in the Southern District of New York, the plaintiffs accused OpenAI of violating the Digital Millennium Copyright Act (DMCA) by stripping copyright management information (CMI) from their articles before using them to train ChatGPT. ... In the ANI v OpenAI case, the Delhi High Court has framed four key issues for adjudication, including whether using copyrighted material for training AI models constitutes infringement and whether Indian courts have jurisdiction over a US-based company. Nath’s view aligns with broader concerns over how existing legal frameworks struggle to keep pace with AI advancements.


Defense strategies to counter escalating hybrid attacks

Threat actor profiling plays a pivotal role in uncovering hybrid operations by going beyond surface-level indicators and examining deeper contextual elements. Profiling involves a thorough analysis of the actor’s history, their strategic objectives, and their operational behaviors across campaigns. For example, understanding the geopolitical implications of a ransomware attack targeting a defense contractor can reveal espionage motives cloaked in financial crime. Profiling allows researchers to differentiate between purely financial motivations and state-sponsored objectives masked as criminal operations. Hybrid actors often leave “behavioral fingerprints” – unique combinations of techniques and infrastructure reuse – that, when analyzed within the context of their history, can expose their true intentions. ... Threat intelligence feeds enriched with historical data can help correlate real-time events with known threat actor profiles. Additionally, implementing deception techniques, such as industry-specific honeypots, can reveal operational objectives and distinguish between actors based on their response to decoys. ... Organizations must adapt by adopting a defense-in-depth strategy that combines proactive threat hunting, continuous monitoring, and incident response preparedness.


4 Cybersecurity Misconceptions to Leave Behind in 2025

Workers need to avoid falling into a false sense of security, and organizations must ensure that they are frequently updating advice and strategies to reduce the likelihood of their employees falling victim. In addition, we found that this confidence doesn’t necessarily translate into action. A notable portion of those surveyed (29%) admit that they don’t report suspicious messages even when they do identify a phishing scam, despite the presence of convenient reporting tools like “report phishing” buttons. ... Our second misconception stems from workers’ sense of helplessness. This kind of cyber apathy can become a dangerous self-fulfilling prophecy if left unaddressed. The key problem is that even if it’s true that information is already online, this isn’t equivalent to being directly under threat, and there are different levels of risk. It’s one thing knowing someone has your home address; knowing they have your front door key in their pocket is quite another. Even if it’s hard to keep all of your data hidden, that doesn’t mean it’s not worth taking steps to keep key information protected. While it can seem impossible to stay safe when so much personal data is publicly available, this should be the impetus to bolster cybersecurity practices, such as not including personal information in passwords.


Real datacenter emissions are a dirty secret

With legislation such as the EU's Corporate Sustainability Reporting Directive (CSRD) now in force, customers and resellers alike are expecting more detailed carbon emissions reporting across all three Scopes from suppliers and vendors, according to Canalys. This expectation of transparency is increasingly important in vendor selection processes because customers need their vendors to share specific numbers to quantify the environmental impact of their cloud usage. "AWS has continued to fall behind its competitors here by not providing Scope 3 emissions data via its Customer Carbon Footprint Tool, which is still unavailable," Caddy claimed. "This issue has frustrated sustainability-focused customers and partners alike for years now, but as companies prepare for CSRD disclosure, this lack of granular emissions disclosure from AWS can create compliance challenges for EU-based AWS customers." We asked Amazon why it doesn't break out the emissions data for AWS separately from its other operations, but while the company confirmed this is so, it declined to offer an explanation. Neither did Microsoft nor Google. In a statement, an AWS spokesperson told us: "We continue to publish a detailed, transparent report of our year-on-year progress decarbonizing our operations, including across our datacenters, in our Sustainability Report. 


5 hot network trends for 2025

AI will generate new levels of network traffic, new requirements for low latency, and new layers of complexity. The saving grace, for network operators, is AIOps – the use of AI to optimize and automate network processes. “The integration of artificial intelligence (AI) into IT operations (ITOps) is becoming indispensable,” says Forrester analyst Carlos Casanova. “AIOps provides real-time contextualization and insights across the IT estate, ensuring that network infrastructure operates at peak efficiency in serving business needs.” ... AIOps can deliver proactive issue resolution, it plays a crucial role in embedding zero trust into networks by detecting and mitigating threats in real time, and it can help network execs reach the Holy Grail of “self-managing, self-healing networks that could adapt to changing conditions and demands with minimal human intervention.” ... Industry veteran Zeus Kerravala predicts that 2025 will be the year that Ethernet becomes the protocol of choice for AI-based networking. “There is currently a holy war regarding InfiniBand versus Ethernet for networking for AI with InfiniBand having taken the early lead,” Kerravala says. Ethernet has seen tremendous advancements over the last few years, and its performance is now on par with InfiniBand, he says, citing a recent test conducted by World Wide Technology. 


Building the Backbone of AI: Why Infrastructure Matters in the Race for Adoption

One of the primary challenges facing businesses when it comes to AI is having the foundational infrastructure to make it work. Depending on the use case, AI can be an incredibly demanding technology. Some algorithmic AI workloads use real-time inference, which will grossly underperform without a direct, high bandwidth, low-latency connection. ... An organization’s path to the cloud is really the central pillar of any successful AI strategy. The sheer scale at which organizations are harvesting and using data means that storing every piece of information on-premises is simply no longer viable. Instead, cloud-based data lakes and warehouses are now commonly used to store data, and having streamlined access to this data is essential. But this shift isn’t just about scale or storage – it’s about capability. AI models, particularly those requiring intensive training, often reside in the cloud, where hyperscalers can offer the power density and GPU capabilities that on-premises data centers typically cannot support. Choosing the right cloud provider in this context is of course vital, but the real game-changer lies not in the who of connectivity, but the how. Relying on the public internet for cloud access creates bottlenecks and risks, with unpredictable routes, variable latency, and compromised security.


Why all developers should adopt a safety-critical mindset

Safety-critical industries don’t just rely on reactive measures; they also invest heavily in proactive defenses. Defensive programming is a key practice here, emphasizing robust input validation, error handling, and preparation for edge cases. This same mindset can be invaluable in non-critical software development. A simple input error could crash a service if not properly handled—building systems with this in mind ensures you’re always anticipating the unexpected. Rigorous testing should also be a norm, and not just unit tests. While unit testing is valuable, it's important to go beyond that, testing real-world edge cases and boundary conditions. Consider fault injection testing, where specific failures are introduced (e.g., dropped packets, corrupted data, or unavailable resources) to observe how the system reacts. These methods complement stress testing under maximum load and simulations of network outages, offering a clearer picture of system resilience. Validating how your software handles external failures will build more confidence in your code. Graceful degradation is another principle worth adopting. If a system does fail, it should fail in a way that’s safe and understandable. For example, an online payment system might temporarily disable credit card processing but allow users to save items in their cart or check account details.


Strengthening Software Supply Chains with Dependency Management

Organizations must prioritize proactive dependency management, high-quality component selection and vigilance against vulnerabilities to mitigate escalating risks. A Software Bill of Materials (SBOM) is an essential tool in this approach, as it offers a comprehensive inventory of all software components, enabling organizations to quickly identify and address vulnerabilities across their dependencies. In fact, projects that implement an SBOM to manage open source software dependencies demonstrate a 264-day reduction in the time taken to fix vulnerabilities compared to those that do not. SBOMs provide a comprehensive list of every component within the software, enabling quicker response times to threats and bolstering overall security. However, despite the rise in SBOM usage, it is not keeping pace with the influx of new components being created, highlighting the need for enhanced automation, tooling and support for open source maintainers. ... This complacency — characterized by a false sense of security — accumulates risks that threaten the integrity of software supply chains. The rise of open source malware further complicates the landscape, as attackers exploit poor dependency management. 

No comments:

Post a Comment