AI & software security
-
EU sets cyber plan for advanced AI
The European Commission has set out an action plan linking advanced AI, cyber defence, model evaluation, and existing EU laws including NIS2, DORA, and the Cyber Resilience Act.
-
Bank warns on frontier AI cyber risk
The Bank of England says frontier AI could increase cyber and operational risks to UK financial stability by accelerating vulnerability discovery and exploitation.
-
ENISA turns CRA readiness towards SMEs
ENISA has published a Cyber Resilience Act maturity model for smaller organisations that manufacture, integrate, or support products with digital elements.
-
EU sets plan for AI cyber risk
The European Commission’s plan links advanced AI, model evaluation, critical infrastructure protection, open-source security, and Europe’s cyber capability base.
-
Microsoft pushes faster Windows patching
Microsoft says AI will increase the pace and volume of Windows security updates, forcing organisations to revisit patch windows, deferrals, and change risk.
-
Dialogflow flaw exposes AI isolation risk
A patched Dialogflow CX flaw shows how AI chatbot infrastructure can turn a single agent permission into conversation exposure and cross-agent risk.
-
Claude Code alert becomes trust test
China’s warning over Claude Code should be treated as a disputed product-security and geopolitical trust story, not a confirmed backdoor case.
-
Fake payment SDKs target developer secrets
Malicious npm and PyPI packages impersonating Paysafe, Skrill, and Neteller SDKs show how payment-brand trust is being used to reach developer credentials and cloud tokens.
-
OpenMandriva case exposes maintainer risk
OpenMandriva says it faced attempted distribution sabotage, putting maintainer privilege, repository control, and recovery discipline back into software supply chain focus.
-
Estonia puts identity around AI agents
Estonia’s plan to create digital identities for AI agents turns machine identity, delegated authority, and auditability into a public-sector governance issue.