AI & software security
-
Epic pauses development after AI testing finds security flaws
Healthcare software provider Epic has paused much of its product development while it addresses security weaknesses identified during testing with Anthropic’s advanced cyber model.
-
UK scientists call for AI accident investigator
Nineteen UK computer scientists have called for an independent authority to investigate serious AI incidents and near misses using a model comparable to accident investigation in aviation.
-
CERT EU flags surge in stolen AI service access
CERT EU says cybercrime markets are increasingly trading stolen access to premium AI services while attackers also hijack cloud servers to run unauthorised AI workloads.
-
ENISA details how CRA cyber reporting now works
ENISA has expanded its guidance for the Cyber Resilience Act’s Single Reporting Platform, detailing how manufacturers and open source stewards submit vulnerability and incident reports across the EU.
-
KYND adds AI visibility to underwriting
KYND has added external AI-technology detection to its cyber-risk platform, giving insurers another source for assessing AI exposure beyond information declared by insured organisations.
-
Thales ties DSPM to data protection
Thales has launched CipherTrust DSPM, combining sensitive-data discovery and risk analysis with encryption, tokenisation, and masking as AI systems gain wider data access.
-
Google limits Argon access to cyber defenders
Google is initially restricting Gemini 4 Argon to trusted cyber defenders as it tests safeguards around a frontier model built for long-horizon technical work.
-
Google charts faster vulnerability exploitation
Google says vulnerability disclosures doubled during 2026 while observed exploitation increased sharply, although attackers still use only a small fraction of disclosed flaws.
-
Cyber budgets rise as continuity lags
PwC says 84% of surveyed security and finance leaders expect cyber budgets to rise, while only 39% report a fully formalised cyber-specific continuity plan.
-
TeamViewer patches five high-severity flaws
TeamViewer has fixed five high-severity vulnerabilities affecting its clients and related services, including weaknesses that can enable remote code execution and local privilege escalation.