Technology

Artificial Intelligence (AI) is rapidly reshaping the landscape of fraud prevention, creating new opportunities for defense as well as new avenues for deception.

Across industries, AI has become a double-edged sword. On one hand, it enables more sophisticated fraud detection, but on the other, it is being weaponized by threat actors to exploit controls, create synthetic identities and launch hyper-realistic attacks.

Fraud prevention is vital in sectors handling high volumes of sensitive transactions and digital identities. In financial services, for example, it's not just about protecting capital - regulatory compliance and customer trust are at stake.

Similar cybersecurity pressures are growing in telecoms and tech industries like SaaS, ecommerce and cloud infrastructure, where threats like SIM swapping, API abuse and synthetic users can cause serious disruption.

Fraud has already shifted from a risk to a core business challenge - with 58 per cent of key decision-makers in large UK businesses now viewing it as a ‘serious threat’, according to a survey conducted in 2024.

The rise of synthetic threats

Synthetic fraud refers to attacks that leverage fabricated data, AI-generated content or manipulated digital identities. These aren’t new concepts, but the capability and accessibility of generative AI tools have dramatically lowered the barrier to entry.

A major threat is the creation of synthetic identities which are combinations of real and fictitious information used to open accounts, bypass Know-Your-Customer (KYC) checks or access services.

Deepfakes are also being used to impersonate executives during video calls or in phishing attempts. One recent example involved attackers using AI to mimic a CEO’s voice and authorize a fraudulent transfer. These tactics are difficult to detect in fast-moving digital environments without advanced, real-time verification methods.

Data silos only exacerbate the problem. In many tech organizations, different departments rely on disconnected tools or platforms. One team may use AI for authentication while another still relies on legacy systems, and it is these blind spots which are easily exploited by AI-driven fraud.

AI as a defense

While AI enables fraud, it also offers powerful tools for defense if implemented strategically. At its best, AI can process vast volumes of data in real time, detect suspicious patterns and adapt as threats evolve. But this depends on effective integration, governance and oversight.

One common weakness lies in fragmented systems. Fraud prevention efforts often operate in silos across compliance, cybersecurity and customer teams. To build true resilience, organizations must align AI strategies across departments. Shared data lakes, or secure APIs, can enable integrated models with a holistic view of user behavior.

Synthetic data, often associated with fraud, can also play a role in defense. Organizations can use anonymized, realistic data to simulate rare fraud scenarios and train models without compromising customer privacy. This approach helps test defenses against edge cases not found in historical data.

Fraud systems must also be adaptive. Static rules and rarely updated models can’t keep pace with AI-powered fraud - real-time, continuously learning systems are now essential. Many companies are adopting behavioral biometrics, where AI monitors how users interact with devices, such as typing rhythm or mouse movement, to detect anomalies, even when credentials appear valid.

Explainability is another cornerstone of responsible AI use and it is essential to understand why a system has flagged or blocked activity. Explainable AI (XAI) frameworks help make decisions transparently, supporting trust and regulatory compliance, ensuring AI is not just effective, but also accountable.

Industry collaboration

AI-enhanced fraud doesn’t respect organizational boundaries, and as a result, cross-industry collaboration is becoming increasingly important. While sectors like financial services have long benefited from information-sharing frameworks like ISACs, similar initiatives are emerging in the broader tech ecosystem.

Cloud providers are beginning to share indicators of compromised credentials or coordinated malicious activity with clients. SaaS and cybersecurity vendors are also forming consortiums and joint research initiatives to accelerate detection and improve response times across the board.

Despite its power, AI is not a silver bullet and organizations which rely solely on automation risk missing subtle or novel fraud techniques. Effective fraud strategies should include regular model audits, scenario testing and red-teaming exercises (where ethical hackers conduct simulated cyberattacks on an organization to test cybersecurity effectiveness).

Human analysts bring domain knowledge and judgement that can refine model performance. Training teams to work alongside AI is key to building synthetic resilience, combining human insight with machine speed and scale.

Resilience is a system, not a feature

As AI transforms both the tools of fraud and the methods of prevention, organizations must redefine resilience. It’s no longer about isolated tools, but about creating a connected, adaptive, and explainable defense ecosystem.

For many organizations, that means integrating AI across business units, embracing synthetic data, prioritizing explainability, and embedding continuous improvement into fraud models. While financial services may have pioneered many of these practices, the broader tech industry now faces the same level of sophistication in fraud, and must respond accordingly.

In this new era, synthetic resilience is not a static end goal but a capability to be constantly cultivated. Those who succeed will not only defend their businesses more effectively but help define the future of secure, AI-enabled digital trust.

We list the best identity management solutions.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro

The landscape of smart data capture software is undergoing a significant transformation, with advancements that can help businesses build long-term resilience against disruptions like trade tariffs, labor shortages, and volatile demand.

No longer confined to handheld computers and mobile devices, the technology is embracing a new batch of hybrid data capture methods that include fixed cameras, drones, and wearables.

If you weren’t familiar with smart data capture, it is the ability to capture data intelligently from barcodes, text, IDs, and objects. It enables real-time decision-making, engagement, and workflow automation at scale across industries such as retail, supply chain, logistics, travel, and healthcare.

The advancements it’s currently experiencing are beyond technological novelties; they are further redefining how businesses operate, driving ROI, enhancing customer experience, and streamlining operational workflows. Let’s explore how:

More than just smartphones

Traditionally, smart data capture relied heavily on smartphones and handheld computers, devices that both captured data and facilitated user action. With advancements in technology, the device landscape is expanding. Wearables like smart glasses and headsets, fixed cameras, drones, and even robots are becoming more commonplace, each with its own value.

This diversification leads to the distinction of devices that purely ‘capture’ data versus those that can ‘act’ on it too. For example, stationary cameras or drones capture data from the real world and then feed it into a system of record to be aggregated with other data.

Other devices — often mobile or wearable — can capture data and empower users to act on that information instantly, such as a store associate who scans a shelf and can instantly be informed of a pricing error on a particular item. Depending on factors such as the frequency of data collected, these devices can allow enterprises to tailor a data capture strategy to their needs.

Practical innovations with real ROI

In a market saturated with emerging technologies, it's easy to get caught up in the hype of the next big thing. However, not all innovations are ready for prime time, and many fail to deliver a tangible return on investment, especially at scale. The key for businesses is to focus on practical, easy-to-implement solutions that enhance workflows rather than disrupt them by leveraging existing technologies and IT infrastructure.

An illustrative example of this evolution is the increasing use of fixed cameras in conjunction with mobile devices for shelf auditing and monitoring in retail environments. Retailers are deploying mobile devices and fixed cameras to monitor shelves in near real-time and identify out-of-stock items, pricing errors, and planogram discrepancies, freeing up store associates’ time and increasing revenue — game-changing capabilities in the current volatile trade environment, which triggers frequent price changes and inventory challenges.

This hybrid shelf management approach allows businesses to scale operations no matter the store format: retailers can easily pilot the solution using their existing mobile devices with minimal upfront investment and assess all the expected ROI and benefits before committing to full-scale implementation.

The combination also enables further operational efficiency, with fixed cameras providing continuous and fully automated shelf monitoring in high-footfall areas, while mobile devices can handle lower-frequency monitoring in less-frequented aisles.

This is how a leading European grocery chain increased revenue by 2% in just six months — an enormous uplift in a tight-margin vertical like grocery.

Multi-device and multi-signal systems

An important aspect of this data capture evolution is the seamless integration of all these various devices and technologies. User interfaces are being developed to facilitate multi-device interactions, ensuring that data captured by one system can be acted upon through another.

For example, fixed cameras might continuously monitor inventory levels, with alerts to replenish specific low-stock items sent directly to a worker's wearable device for immediate and hands-free action.

And speaking of hands-free operation: gesture recognition and voice input are also becoming increasingly important, especially for wearable devices lacking traditional touchscreens. Advancing these technologies would enable workers to interact with items naturally and efficiently.

Adaptive user interfaces also play a vital role, ensuring consistent experiences across different devices and form factors. Whether using a smartphone, tablet, or digital eyewear, the user interface should adapt to provide the necessary functionality without a steep learning curve; otherwise, it may negatively impact the adoption rate of the data capture solution.

Recognizing the benefits, a large US grocer implemented a pre-built adaptive UI to enable top-performing scanning capabilities on existing apps to 100 stores in just 90 days.

The co-pilot system

As the volume of data increases, so does the potential for information overload. In some cases, systems can generate thousands of alerts daily, overwhelming staff and hindering productivity. To combat this, businesses are adopting so-called co-pilot systems — a combination of devices and advanced smart data capture that can guide workers to prioritize ROI-optimizing tasks.

This combination leverages machine learning to analyze sales numbers, inventory levels, and other critical metrics, providing frontline workers with actionable insights. By focusing on high-priority tasks, employees can work more efficiently without sifting through endless lists of alerts.

Preparing for the future

As the smart data capture landscape continues to evolve and disruption becomes the “new normal”, businesses must ensure their technology stacks are flexible, adaptable, and scalable.

Supporting various devices, integrating multiple data signals, and providing clear task prioritization are essential for staying competitive in an increasingly complex, changeable and data-driven market.

By embracing hybrid smart data capture device strategies, businesses can optimize processes, enhance user experiences, and make informed decisions based on real-time data.

The convergence of mobile devices, fixed cameras, wearables, drones, and advanced user interfaces represents not just an evolution in technology but a revolution in how businesses operate. And in a world where data is king, those who capture it effectively — and act on it intelligently — will lock in higher margins today and lead the way tomorrow.

We've listed the best ERP software.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro

Google Gemini introduced a new feature aimed at education called Guided Learning this month. The idea is to teach you something through question-centered conversation instead of a lecture.

When you ask it to teach you something, it breaks the topic down and starts asking you questions about it. Based on your answers, it explains more details and asks another question. The feature provides visuals, quizzes, and even embeds YouTube videos to help you absorb knowledge.

As a test, I asked Gemini's Socratic tutor to teach me all about cheese. It started by asking me about what I think is in cheese, clarifying my somewhat vague answer with more details, and then asking if I knew how those ingredients become cheese. Soon, I was in a full-blown cheese seminar. For every answer I gave, Gemini came back with more details or, in a gentle way, told me I was wrong.

The AI then got into cheese history. It framed the history as a story of traveling herders, clay pots, ancient salt, and Egyptian tombs with cheese residue. It showed a visual timeline and said, “Which of these surprises you most?” I said the tombs did, and it said, “Right? They found cheese in a tomb and it had survived.” Which is horrifying and also makes me respect cheese on a deeper level.

In about 15 minutes, I knew all about curds and whey, the history of a few regional cheese traditions, and even how to pick out the best examples of different cheeses. I could see photos in some cases and a video tour of a cellar full of expensive wheels of cheese in France. The AI quizzed me when I asked it to make sure I was getting it, and I scored a ten out of ten.

(Image credit: Gemini screenshots)Cheesemonger AI

It didn’t feel like studying, exactly. More like falling into a conversation where the other person knows everything about dairy and is excited to bring you along for the ride. After learning about casein micelles. starter cultures, and cutting the curd, Gemini asked me if I wanted to learn how to make cheese.

I said sure, and it guided me through the process of making ricotta, including pictures to help show what it should look like at each step.

(Image credit: Gemini screenshots)

By the time I was done with that part of the conversation, I felt like I’d taken a mini‑course in cheesemaking. I'm not sure I am ready to fill an entire cheeseboard or age a wheel of gruyère in my basement.

Still, I think making ricotta or maybe paneer would be a fun activity in the next few weeks. And I can show off a mild, wobbly ball of dairy pride thanks to learning from questioning, and, as it were, being guided to an education.

You might also like

• Google Gemini 2.5 Flash promises to be your favorite AI chatbot, but how does it compare to ChatGPT 4o?
• 4 ways ChatGPT Tasks can help you take control of your life – trust me it's my favorite AI tool of 2025 so far
• Google plans to get its AI to write your emails for you

Here are the answers for The New York Times Mini Crossword for Aug. 12