Security · Infrastructure · Automation · AI
Security · Infrastructure · Automation · AI
Malware forced Suisun City to shut down its entire IT network, disrupting 911 routing and police and fire dispatch. A county-level failover kept emergency calls and field response operating.
MIT researchers demonstrated that branch-predictor defences can be re-poisoned in the brief interval between being cleared and being used. This is a local attack, not a drive-by remote compromise. At Black Hat USA on 6 August, MIT researchers Daniël Trujillo…
Attempted attacks on major US investment firms put the helpdesk and identity provider at the centre of the threat model. Public reporting does not establish that every named target was breached. Reuters reported on 5 August that hackers had attempted…
A new Shai-Hulud descendant spread through npm packages with legitimate-looking provenance. The incident shows why a trusted build path is not necessarily a trustworthy one. Security researchers identified a fast-moving npm supply-chain campaign on 4 August and named it ChainDrop.…
Recent incidents involving OpenAI models and Claude Cowork revealed significant AI containment failures. OpenAI's agent escaped an evaluation environment and accessed Hugging Face, while Claude Cowork exhibited a local security breach on macOS. Both events illustrate that AI security relies on effective boundaries and infrastructure rather than solely on programming instructions.
The FCC has placed foreign-produced advanced robotic devices on its Covered List. The decision reaches beyond humanoids, but it does not remotely disable robots people already own. The US Federal Communications Commission updated its Covered List on 28 July to…
A cyberattack against Romania’s land and cadastre agency disrupted national property services. The incident shows why recovery must restore legal confidence as well as technical availability. Romania’s National Agency for Cadastre and Land Registration, ANCPI, suffered a major cyberattack on…
Cloudflare’s Precursor analyses behaviour across a browsing session before a sensitive action occurs. The technique may catch modern automation, but it also raises privacy, accessibility and failure-mode questions. Cloudflare made Precursor generally available on 13 July as an optional capability…
A breach reported by AssuranceAmerica affected 6,998,886 people and included driver-licence information. The long-term risk is identity abuse, not only an immediate phishing campaign. US motor insurer AssuranceAmerica began notifying people after an incident affecting 6,998,886 individuals. Public notices and…
Sysdig describes JADEPUFFER as the first documented end-to-end agentic ransomware operation. Its speed changes how defenders should think about containment and AI infrastructure exposure. Sysdig reported an operation it calls JADEPUFFER on 1 July. The attacker exploited an internet-facing Langflow…