My Courses
-
Researchers have outlined a Windows EPM poisoning exploit sequence that can result in escalated privileges within a domain.
Cybersecurity researchers have unveiled critical findings regarding a now-patched security vulnerability in Microsoft’s Windows Remote Procedure Call (RPC) communication protocol. This vulnerability, tracked as CVE-2025-49760 with a CVSS score of 3.5, has been identified as a Windows Storage spoofing bug. It was addressed in July 2025 during Microsoft’s monthly Patch Tuesday update. SafeBreach researcher Ron…
-
Scientists have discovered vulnerabilities in GPT-5 that allow for jailbreaks and zero-click AI agent attacks, which threaten the security of cloud and IoT systems.
Cybersecurity researchers have discovered a jailbreak technique that circumvents the ethical guardrails implemented by OpenAI in its latest large language model, GPT-5, enabling the generation of illicit instructions. The generative artificial intelligence security platform, NeuralTrust, reported that it utilised a known method called Echo Chamber in conjunction with narrative-driven steering to manipulate the model into…
-
A vulnerability in Lenovo webcams that run on Linux can be exploited remotely, allowing for potential BadUSB attacks.
Cybersecurity researchers have identified vulnerabilities in specific Lenovo webcam models that could enable them to function as BadUSB attack devices. This discovery allows remote attackers to covertly inject keystrokes and launch attacks that operate independently of the host operating system. The vulnerabilities, referred to as BadCam, were detailed in a report by Eclypsium researchers Paul…
-
Complimentary Wi-Fi makes buses susceptible to hacking from remote locations.
Researchers have uncovered significant vulnerabilities in the onboard and remote systems of some buses, revealing how these flaws can be exploited by hackers for malicious purposes such as tracking, control, and spying. The study highlights that the provision of free Wi-Fi on public transport systems, while convenient for passengers, inadvertently opens the door to cyber…
-
RubyGems and PyPI have both experienced attacks involving harmful packages that are designed to steal credentials and cryptocurrency
A recent investigation has revealed a new set of 60 malicious packages targeting the RubyGems ecosystem. These packages masquerade as harmless automation tools for social media, blogging, and messaging services, with the intent to steal user credentials and potentially resell them on dark web forums like Russian Market. The software supply chain security company Socket…
-
Exploring the emerging trends in cyber deception threats of 2025, from counterfeit CAPTCHAs to Remote Access Trojans (RATs).
Cybercriminals are becoming increasingly adept at deception, as highlighted in a recent LevelBlue report. Attackers are utilising social engineering techniques and legitimate tools to navigate through environments undetected. The report reveals a significant increase in the number of customers affected by security incidents, which nearly tripled from 6 per cent in late 2024 to 17…
-
Fraud prevention measures do not ensure that consumers will have confidence.
According to Experian, over a third of companies are leveraging AI, including generative AI, to combat fraud as threats become increasingly sophisticated. In response, businesses are ramping up their investments, with more than half adopting new analytics and developing AI models to improve customer decision-making. However, AI presents a paradox; while it enhances the speed…
-
Red Teams have successfully bypassed GPT-5’s safeguards, cautioning that it has become ‘almost impractical’ for business use.
Researchers have revealed that multi-turn “storytelling” attacks can effectively bypass the prompt-level filters of GPT-5, highlighting significant systemic weaknesses in its security measures. These findings come from Red Teams that have successfully executed jailbreaks on GPT-5 with relative ease, raising concerns about the model’s reliability in enterprise settings. The ability of these attacks to manipulate…
-
AI technologies utilised for Brazilian phishing scam
Cybersecurity researchers have highlighted a new campaign leveraging legitimate generative artificial intelligence (AI)-powered website building tools, such as DeepSite AI and BlackBox AI, to create replica phishing pages that mimic Brazilian government agencies. This financially motivated activity involves the development of lookalike sites imitating Brazil’s State Department of Traffic and Ministry of Education, which deceive…
-
Help Desks in jeopardy: Scattered Spider highlights an often ignored threat.
As attackers increasingly target help desks and identity systems, traditional security perimeters are proving inadequate against agile, socially-engineered threats. The rise of sophisticated tactics employed by cybercriminals highlights the vulnerabilities within these critical support functions. Organisations must recognise that help desks are often the first line of defence and are now prime targets for exploitation.…