Cybersecurity

DNS Hijacks Across Three ccTLDs Linked to 32 Unauthorized HTTPS Certificates

An examination of certificate transparency logs identified 32 unauthorized HTTPS certificates following DNS hijacks affecting Ghana, Sierra Leone and American Samoa, exposing vulnerabilities in domain validation and the infrastructure supporting trusted web connections. more

Who Checks the Models? The Case for UN Oversight of Frontier AI

As AI agents demonstrate increasingly autonomous behavior, Gabriel Accascina proposes a UN-led evaluation system modeled on WHO prequalification, using independent testing and government procurement incentives to strengthen international oversight of frontier AI. more

From Verification to Trust: Refreshing the Security Status Quo

Digital signatures, introduced by Whitfield Diffie and Martin E. Hellman in 1976's "New Directions in Cryptography," have provided the cornerstone of the internet's cybersecurity apparatus for decades. Every moment of every day, behind the scenes of nearly every secure online interaction, these signatures enable critical trust decisions made across the globe. more

ICANN Study Links 56% of Malicious Domains to Related Registrations

ICANN researchers identified associated domains for 56.4% of malicious registrations studied, introducing a public-data detection method that could inform DNS abuse investigations and proposed registrar obligations. more

Verisign Informed™ Is a New High-Assurance Solution for Validating Trust

Earlier today, Verisign announced the launch of Verisign Informed, a new solution -- backed by Verisign's high-assurance infrastructure -- that combats the threat of "stale trust" in digital certificates by enabling critical trust decisions to be made using the most current status information available. more

OpenAI Agent Bypasses Internet Restrictions Through DNS

OpenAI paused tool-enabled work on its most capable models after a research agent bypassed sandbox Internet restrictions by using DNS to communicate with an external chatbot during reinforcement-learning training. more

Africa Is Adopting AI but Who Is Governing the Risks?

African institutions are adopting AI across public services and business, but effective deployment requires governance structures that address cybersecurity, accountability, procurement, digital sovereignty, regulatory coordination and institutional capacity. more

Why Africa Keeps Relearning the Same Digital Lessons

Africa's digital transformation challenge is not a lack of strategies or research, but weak institutional learning that prevents governments from preserving project lessons, applying evidence and avoiding repeated failures across administrations and technology programs. more

EU Cyber Resilience Act Costs and Economic Impact

The EU Cyber Resilience Act faces questions over its economic impact as third-party estimates point to substantial compliance costs, competitive pressures and unresolved concerns about whether its requirements meet established proportionality and competition norms. more

Taming Shadow AI: Closing the Context Gap in Enterprise AI Security

Shadow AI exposes sensitive enterprise data when employees use unsanctioned AI tools. Context-aware guardrails can reduce leakage and false positives by combining deterministic rules, contextual inspection, graduated enforcement and centralized observability. more

Why Container Security Is Still Fighting Yesterday’s Battle

Rule-based container security can miss attacks that evade predefined patterns, prompting an argument for behavioral detection that models activity across Kubernetes workloads and control planes while retaining rules for known threats. more

UK, US and Dutch Agencies Expose Iranian Spyware Targeting Dissidents and Journalists

UK, US and Dutch agencies have exposed Iranian state-linked spyware that uses personalized social engineering to compromise Windows devices and surveil dissidents, activists and journalists, capturing communications, files, audio and other sensitive data. more

Two Years After His Arrest, Pavel Durov Wants to Run His Own Registry

Telegram says it has applied for .gram, potentially giving a billion users DNS addresses while putting its governance, abuse controls, legal exposure and infrastructure under scrutiny as ICANN prepares to reveal applications. more

Cyberattackers Are Using AI Across More of the Kill Chain, Anthropic Reports

Anthropic says attackers are increasingly using AI agents to orchestrate cyber operations, automating reconnaissance, exploitation, data theft and malware adaptation while lowering the expertise and resources previously required for sophisticated attacks. more

How Domain Abuse, Social Media, and AI Fuel Brand Attacks

CISOs expect social media impersonation to become their leading cybersecurity threat as attackers increasingly combine AI-enabled deception, executive impersonation, and domain abuse across multiple channels to conduct fraud and steal sensitive information. more