AppSec stories
UK security leaders are warning that AI-generated code is outpacing controls, with a quarter already seeing incidents from flaws it introduced.
Cybersecurity experts warn single-person approvals are now vulnerable after an AI agent used fabricated identities to slip malicious code past checks.
As attackers use AI to speed up phishing and malware, companies are being told that multi-factor authentication and patching matter more than ever.
Security teams could cut exposure faster as the model helps rank exploitable flaws and apply temporary defences before vendor fixes arrive.
Regulated teams can now keep AI processing inside GitLab Dedicated, with new secret handling, spending caps and security fixes added in 19.3.
Basic security lapses are leaving web apps exposed, with Barracuda saying routine misconfigurations account for most of 20 flaws per site.
Misconfigured models are giving attackers a fresh route into cloud systems, raising the risk of data theft and service compromise.
The new tool aims to catch Bitcoin software flaws between formal audits after a regression led to more than USD $116 million stolen.
Developers can now scan C code earlier in the process, as Endor Labs says its buildless AI SAST found 96 of 102 known bugs in tests.
Organisations using Microsoft automation can now keep credentials out of scripts, reducing the risk of exposed secrets in cloud workflows.
In two days, the system uncovered more than 100 critical bugs in stolen code repositories, outpacing manual review and aiding incident response.
Unauthorised access could let attackers send arbitrary commands to spacecraft and instruments via NASA's AIT-GUI console, now fixed in version 2.5.2.
Security teams are drowning in AI-generated bug reports, and 11 early users are already testing a system that filters the noise.
Verified access to Anthropic's Claude models should sharpen ArmorCode's exploitability scoring as security teams race to cut alert noise.
Security teams should treat AI patching with caution after 53.9% of 6,080 model-generated fixes failed or introduced new flaws.
Hong Kong saw a record 15,877 cyber incidents in 2025 as AI speeds attacks and shortens the window to exploit software flaws.
Rising use of AI coding tools is widening software supply-chain risks for businesses across the Middle East, Türkiye and Africa.
The new tools aim to cut the gap between finding a flaw and fixing it from weeks to hours as attacks accelerate.
Stolen credentials can become an operational foothold within hours, leaving annual assessments too slow to catch the real attack paths.
Accenture's double win underlines Wiz's push to deepen its Australian channel as customers seek help securing cloud and AI projects.
.webp)
