News

Experiments involving autonomous AI agents coordinating, sharing code and acting without human supervision have captured plenty of attention in recent weeks. Platforms such as Moltbook and tools like Clawdbot have been framed as glimpses into a future where AI systems organise themselves in the wild. For security leaders, however, the fascination quickly fades once the underlying mechanics are examined. According to Salt Security, these experiments are less about artificial intelligence behaving unpredictably and more about...

Modern applications aren't built from scratch. They’re assembled. When you look at most typical enterprise codebases, you’ll see that they contain a large proportion of open source elements, many of which are pulled from public repositories and bolted together with their own proprietary logic.  From a development perspective, this approach makes a lot of sense. Why spend months writing out your own authentication libraries or data parsing various functions when you can just plug and...

A recently discovered online database containing 149 million stolen usernames and passwords has been taken offline after being identified by security researcher Jeremiah Fowler. While the exposure has now been addressed, the scale and nature of the data involved underline a far deeper and ongoing cybersecurity challenge: the industrialisation of credential theft through infostealing malware. According to Fowler’s findings, the database held login details for a vast range of services, including major email providers, social...

Bring your own device (BYOD) programs are no longer a niche policy reserved for a few departments. Today, nearly every organisation allows team members to use their personal hardware for work purposes. However, sweeping adoption has resulted in a tricky security reality. Verizon’s 2025 Mobile Security Index found that 70% of mobile devices impacted by cyber attacks are personal rather than corporate-issued, which means the riskiest endpoints are often the ones that the business doesn’t...

BreachForums, one of the most well-known English-language cybercrime forums, has reportedly suffered a data breach, exposing user information after the site was taken offline once again. As reported by The Register, a database linked to the forum was leaked online, potentially revealing account details, private messages and metadata on close to 325,000 accounts. However, security researchers caution that while the leak may attract attention, its intelligence value and authenticity remain uncertain. Michael Tigges, Senior Security...

Pharmaceutical research teams face a constant challenge. They need fast, accurate documentation, but the writing process can slow things down. Research teams have to spend countless hours drafting clinical reports for regulatory agencies and double-checking details, buried under mountains of manual work. Artificial intelligence (AI) is changing all this. It can cut the time spent on regulatory documentation by up to 40%, a case study on a specialised AI tool revealed. For busy research departments,...

The philosophy of data collection in cybersecurity has undergone a radical transformation over the last two years. For over a decade, organisations operated under the assumption that accumulating vast amounts of user information provided a competitive advantage and a foundation for future analytics.  However, as we move through 2026, this "collect everything" mindset has morphed into a significant liability, with CISOs now viewing data repositories as toxic assets that increase the attack surface. The shift...

As automated attack networks grow larger and more sophisticated, security teams are struggling to keep pace with a surge in malicious bot activity that is reshaping the DDoS threat landscape In December 2025, Solana experienced one of the largest DDoS attacks in history, with traffic peaking at 6 Tbps. Although the attack continued over more than a week, Solana reported zero network down time. Had the attack succeeded, it could have scammed everyday retail investors...

Government data has been stolen in a cyberattack, though officials say the risk to individuals remains low, according to a UK minister. The incident has prompted an ongoing investigation and renewed warnings from cybersecurity experts about the long-term risks of state-linked digital espionage. Trade Minister Chris Bryant confirmed the breach in an interview with BBC Breakfast, saying officials moved quickly once the issue was identified. “An investigation is ongoing,” Bryant said, adding that the security...

UK-based AI safety and governance company CultureAI has been named as one of the participants in Microsoft’s newly launched Agentic Launchpad, a technology accelerator aimed at supporting startups working on advanced AI systems. The inclusion marks a milestone for CultureAI’s growth and signals broader industry interest in integrating AI safety and usage control into emerging autonomous AI ecosystems. The Agentic Launchpad is a collaborative programme from Microsoft, NVIDIA, and WeTransact designed to support software companies...