Cybersecurity

Open-Weight AI: Open to Whom?

The push for open-weight AI promises broader access and competition, but true openness requires more than releasing model weights. Compute, infrastructure, training data, security, and control ultimately determine who can meaningfully benefit from open AI. more

AI Duties Are Being Written but the Protocol Layer Is Missing

Recent AI containment failures expose a governance gap: harmful model behavior is often visible at the network boundary but poorly monitored. Adapting proven internet protocols could give governments and companies practical tools to enforce accountability. more

Trump Enlists Private Firms for Offensive Cyber Operations

The Trump administration will allow vetted American companies to conduct surveillance and offensive cyber operations against foreign criminal groups, extending private-sector capabilities into cybercrime enforcement while keeping missions subject to federal approval, oversight and legal safeguards. more

DNS Security Gets Fixed When Someone Notices - Not Before

A study of 309 universities, museums and sports brands finds DNS security follows visible threats rather than technical risk, leaving organisations well protected against familiar attacks while quieter weaknesses in critical infrastructure persist unnoticed. more

Phantom Squatting: When LLMs’ Hallucination Becomes an Attacker’s Best Friend

As AI reshapes cybersecurity, attackers are exploiting a fundamental weakness in large language models. Phantom squatting turns hallucinated domain names into trusted attack vectors, creating a new class of DNS abuse that defenders cannot solve by eliminating hallucinations alone. more

Microsoft Launches AI Cybersecurity Model to Boost MDASH Performance and Cut Costs

Microsoft has unveiled a specialist cybersecurity AI model powering MDASH, claiming higher benchmark performance and lower costs while launching Project Perception, an agentic security platform designed to help defenders counter increasingly automated cyber threats. more

The EU Cyber Resilience Act Regime: A Failure to Know Your Limitations

A cybersecurity historian argues the EU Cyber Resilience Act overreaches through unworkable mandates, sprawling jurisdiction and outdated assumptions, urging a streamlined successor that embraces AI, existing standards and practical enforcement instead of bureaucratic complexity today. more

Who Registers the Domain That Steals Your Data? The Registrar Accountability Gap in India’s DNS Abuse Crisis

India's DNS abuse crisis exposes a registrar accountability gap: malicious domains remain online while vulnerable users bear the cost. Concentrated abuse patterns suggest stronger oversight and faster mitigation could significantly reduce predictable harms nationwide today. more

The Question Isn’t Whether the Harm Is Real - It’s Who Should Act

Measuring online abuse can reveal its scale, but not who should intervene. Effective policy must distinguish harm from contractual responsibility, identify the actors best placed to act, and target remedies where they can work effectively. more

ZEGO’s Cyberattack Pushes Bavarian Textile Firm Into Insolvency

A cyberattack forced Bavarian textile specialist ZEGO into insolvency after a six-week production shutdown. The company is pursuing restructuring efforts to preserve jobs, customer relationships and its future amid growing cyber risks. more

FBI Warns of Russian Cyber Campaign Targeting Vulnerable Routers

The FBI and international partners warn that Russia's FSB Center 16 is exploiting vulnerable routers worldwide, highlighting persistent weaknesses in network security and urging organizations to adopt stronger protections to safeguard critical infrastructure against intrusion. more

UK Unveils AI Cyber Shield to Counter Machine-Speed Digital Threats

Britain is developing Cyber Shield, an AI-powered national cyber defense program designed to detect threats, automate responses and protect critical infrastructure as increasingly sophisticated attacks challenge conventional security and strain existing defensive capabilities. more

Spain to Require Four Hours of Mobile Service During Power Blackouts

Spain will require telecom operators to keep mobile networks running for up to four hours during power outages, introducing phased resilience standards, stronger emergency communications safeguards and stricter backup requirements for critical digital infrastructure. more

.PK ccTLD Governance Issues and Pakistan’s Digital Future

Pakistan's .pk domain has long been controlled by a private company abroad, raising concerns over digital sovereignty, cybersecurity and accountability. Repeated breaches, offshore infrastructure and weak governance have left a critical national asset exposed and contested. more

When AI Writes the Scam: How Artificial Intelligence Is Making DNS Abuse Harder to Detect

Artificial intelligence is transforming phishing and DNS abuse, erasing the linguistic clues that once exposed scams. As attacks become personalised, automated and multilingual, governance frameworks are struggling to keep pace with a rapidly expanding threat surface. more