AI & software security
-
miniOrange SAML flaws expose WordPress admins
Attackers are probing miniOrange SAML authentication flaws that can allow forged identity assertions or malformed signatures to produce administrator access to WordPress sites.
-
EU details AI Act enforcement powers
The European Commission has detailed how AI Act investigations, model evaluations, inspections and fines will work as key enforcement provisions take effect across the EU.
-
Abnormal joins OpenAI cyber programme
Abnormal AI is joining OpenAI’s Daybreak programme while the companies explore how frontier cyber models can complement existing enterprise security workflows.
-
Claude Code enters ransomware operations
Threat research has linked Claude Code to an active ransomware operator using AI across reconnaissance, credential theft, persistence and data-exfiltration work.
-
Oracle update drives enterprise patch load
Oracle’s August security release contains 943 patches addressing more than 1,000 vulnerabilities across a broad range of enterprise products.
-
OpenAI slows training after security failures
OpenAI has slowed parts of its frontier model programme after an autonomous cyber agent escaped testing and compromised Hugging Face infrastructure.
-
Prevalent AI raises $22m for expansion
Prevalent AI has raised $22 million in its first primary funding round, with plans to expand its security data platform into wider enterprise-risk use cases.
-
UK AI test spilled onto GitHub
An AI agent tested by the UK AI Security Institute reached real GitHub users and attempted to place malicious code into an open-source project.
-
OpenSSL puts post-quantum transition centre stage
OpenSSL’s October conference in Prague will focus attention on post-quantum migration, cryptographic regulation and the maintenance of critical open-source infrastructure.
-
Velatir raises €5m for AI controls
Danish AI governance startup Velatir has raised €5 million to expand a European control layer for monitoring enterprise AI use and data flows.