MatrixPDF reshapes ordinary files into covert lures for unsuspecting victimsSpamGPT campaigns could massively scale the reach of hidden payloadsHarmless documents morph into convincing traps carrying silent, malicious codeResearchers are drawing attention to a new toolkit called MatrixPDF that can turn ordinary documents into delivery vehicles for malware and phishing campaigns.Varonis research found the toolkit modifies existing PDF files to include deceptive prompts, overlays, and scripts, making them appear routine while concealing hidden threats.Experts have warned that pairing this with large-scale phishing engines like SpamGPT could multiply the reach and effectiveness of such campaigns.Fake “Secure Document” promptsMatrixPDF relies on the fact PDF files are widely trusted, often slipping through email filters and opening directly in services like Gmail without raising suspicion.Attackers can load a legitimate document into the builder and insert malicious actions, such as fake “Secure Document” prompts or blurred overlays that prompt a user to click.These interactions can trigger redirections to external sites or even the automatic retrieval of files that compromise the system.One attack method promoted with the toolkit involves phishing link redirection.A PDF which looks genuine can bypass a secure email by containing no embedded ransomware but instead a link or button that directs the user to a payload site.Because the malicious action only occurs when the user clicks, the PDF itself appears safe during automated scans.Once redirected, the victim may unknowingly download a compromised executable, convinced it is part of a secure process.The second approach leverages PDF-embedded JavaScript. In this scenario, the file executes a script as soon as the document opens or when the user interacts with it.This script can attempt to connect to an attacker’s server through a shortened domain, creating the impression of a legitimate resource.When confronted with a security dialog, many users may click “Allow,” not realizing they are enabling the download of malware.At that point, the attack becomes a drive-by download, with the harmful payload installed under the guise of accessing a secure file.The attack exploits user trust with routine phrases like “document is trying to connect...” which usually signals nothing more than a required step to access information.This reliance on social engineering means attackers do not need new exploits; they simply weaponize the credibility of the PDF format itself.In an exclusive exchange with TechRadar Pro, lead researcher Daniel Kelley said, “MatrixPDF and SpamGPT could complement each other in an attack scenario… with one generating malicious PDFs and the other distributing them at scale.”“Combining tools like these allows attackers to scale their operations while maintaining a level of customization and sophistication.”The concern is less about a single exploit and more about how trusted file formats can be systematically reshaped into widespread delivery mechanisms for fraud and malware.AI-based email security is a viable countermeasure because it can analyze attachments beyond signatures, looking for unusual structures, hidden links, or blurred content.By simulating user interactions in a controlled environment, it can expose hidden redirects and scripts before the file ever reaches an inbox.While such defenses improve detection rates, the persistence of these tactics demonstrates the constant adaptation of cybercriminal tools.You might also likeNail the basics with the best firewalls available nowTake a look at our guide to the best authenticator appWe've rounded up the best password managers

Figma releases wide-ranging AI surveyAdoption and roll-out increasing for design and developmentBut users still think it’s not as useful as it could be

As artificial intelligence continues its path to world domination, Figma has unveiled a new survey that seeks to explore how designers and developers are deploying AI tools in the field.The findings are a bit of a mixed bag, showing that while AI is becoming an integral part of the workflow, it’s not always living up to the promise.The survey follows on from news that Figma, alongside Adobe and Canva, has integrated OpenAI’s ‘gpt-image-1’ model into its design platform.

What users really think about AI

85%. That’s the headline figure from the survey, which states “85% of respondents say AI will be essential to future success in their role.” But how that success is achieved, or what it even looks like, remains unclear right now.The company has explored how businesses are building AI-powered products for the market, as well as uncovering how design and dev teams are using the tools in day-to-day workflows. Either way, the results show that despite embracing the technology, right now it’s failing to live up to frankly over-hyped expectations.According to Figma, one in three say they’ve shipped AI-powered products - a 50% increase since last year’s survey. This includes a wide range of uses, including analytics, document interpretation, and product strategy and marketing growth.However, designers and developers admitted that “many AI projects still lack clarity in purpose,” with Figma pointing to nebulously defined goals like AI experimentation and CX improvements offered by respondents. This, the company claims, makes it difficult to measure the true impact of artificial intelligence.The gulf between reality and potential can also be seen in how designers and developers are using AI. Because, while 78% believe the use of AI makes their workflows more efficient, just 58% reckon it improves the quality of the work. Fewer than half of those surveyed, says Figma, “made them better at their job.”Figma also notes that despite 83% of those surveyed think learning to work with AI is essential to future success, those who believe it will have a “significant impact” on reaching company goals plummets to only 27%. Users who say AI will be “transformational” linger at 15%, exactly where it was in last year’s survey.In other words, the company says, belief in the potential is high, but expectations, while stabilizing, are low due to experiencing the true limitations of current artificial intelligence technology.Despite the general scepticism over just how useful AI is right now, there is some optimism in the design and dev world. Agentic AI is now the fastest-growing category, doubling year on year - although it’s yet to unseat text generation as the biggest project type.Usage is also increasing, particularly among developers. Where just 69% of designers say they’re satisfied with AI tools and 54% of them feel the quality of their work improves when using AI, these figures jump to 83% and 67% for those working in development.However, Figma points out that the disparity is likely down to how artificial intelligence is used across different projects, with designers chiefly using the tools for the likes of asset creation (31%) compared to core development responsibilities, such as code generation, used by 59% of developers.So, all in all, results are mixed right now - how that will change as AI continues to improve will be one to watch.Figma surveyed 2500 developers and designers globally between January 8 and January 30, 2025. You can get the full report here.

You might also like

We took a look at the best mockup software for small businesses‘It's like magic and everything just works’: We spoke to Adobe’s AI maestro to find out what’s new with Firefly and how it levels up creativity for allWe've rounded up the best AI writer tools.