AI fills the gap as cybersecurity budgets shrink, and other cybersecurity news

Despite the number of cyberattacks continuing to grow rapidly, cybersecurity budgets' growth rate has slowed markedly, and AI is stepping into the breach.

Industry analysts Ians reports a decline from 17% in 2022 to just 4% in 2025 as economic and geopolitical pressures force organizations to do more with less. Relative to companies' revenues, security budgets are stagnating, with nearly half of security professionals stating that revenues grew faster than the security budget.

Along with delays in rolling out new security initiatives and making difficult trade-off between investments, staff shortages are one of the biggest concerns. Hiring of security staff has slowed and nearly 9 in 10 heads of information security report being understaffed.

With acute talent scarcity, not only are cybersecurity professionals hard to come by, but they are becoming too expensive for many companies, as budgets are not keeping pace with security threats and organizational growth.

In response, many organizations are accelerating their adoption of AI-driven tools. Routine tasks such as threat detection and alert triage are a particular focus for automation using agentic AI. These repetitive tasks are where AI excels and can deliver unmatched consistency. Not only does this generate productivity advantages, it can also help professionals develop their skills and work more effectively.

With millions of students the world over returning to school, the education sector also has to contend with being the most targeted sector for cyber attacks. New research from Check Point shows that education has seen an unprecedented amount of cybercrime between January and July 2025.

In that period, it experienced a 41% increase in attacks compared to the previous year, with an average of 4,356 attacks per week.

With close to 8,000 attacks per week per establishment on average, schools, colleges and universities in Asia Pacific were the most affected, while North America (+67%) and Africa (+56%) registered the biggest year-on-year increases.

The surge in attacks is linked to seasonal periods, such as the back-to-school season, during which cybercriminals use tactics like phishing campaigns that mimic university login pages to steal credentials and payment update scams to get sensitive information.

The education sector is prone to attacks due to a range of vulnerabilities, Check Point says: limited cybersecurity budgets, a distributed user base and a high reliance on online platforms with access to sensitive research and student data.

Following a major ransomware attack in April, Marks & Spencer (M&S) has resumed its click-and-collect service after a 15-week suspension. The attack, which disrupted online ordering and in-store operations, is estimated to cost the retailer around $400 million in annual profits.

A single change in an AI engine's reasoning process could have drastic consequences for applications like autonomous vehicles, facial authentication, or medical imaging. Researchers warn that this 'bit flip' is a feasible risk that security professionals must now mitigate.

Anthropic, the company behind the chatbot Claude, has warned that its AI technology is being 'weaponized' by hackers to write malicious code. In one case, the AI was used to write code that helped breach at least 17 organizations' security. It even helped the hackers make tactical decisions on which data to steal and suggested ransom amounts.

Data breaches are mushrooming and more than a third are linked to third parties -people you buy from or partners you work with as an individual or organization. Jon Jacobson, CEO and Co-Founder of Omnisient, makes the case for privacy enhancing technologies (PETs) to enable organizations to jointly use data without risk of exposure.

Some may challenge globalization, but cybercriminals are revelling in it. Expanding their networks across the world's regions means they are hard to target — but not impossible. Sean Doyle and Natalia Umansky from the World Economic Forum's Cybercrime Atlas Initiative report on the dismantling of 25 cryptocurrency mining centres in Angola.

Deepfakes can be funny - but also manipulative and dangerous. Denmark is now taking action by strengthening its copyright laws; people affected by deepfake content can now request its removal. Artists and their heirs will be able to claim compensation if their image is being used without authorization.