News
-
Cohere AI Terrarium sandbox flaw can let attackers run code as root
A critical flaw in Cohere AI’s Terrarium Python sandbox could allow arbitrary code execution as root, with CERT/CC warning that the bug may let attackers escape the sandbox and reach host or container resources.
-
French government agency confirms data breach after hacker claims 19 million records
France Titres, the French agency that issues identity and registration documents, said a security incident may have exposed account data. A hacker later claimed to be selling up to 19 million records.
-
NGate malware campaign targets Brazil through trojanized HandyPay app
Researchers found a new NGate Android malware campaign targeting Brazil since around November 2025. The trojanized HandyPay app can relay NFC payment data, capture PINs and help thieves carry out fraudulent ATM withdrawals.
-
Google patches Antigravity IDE flaw that could enable code execution
Google has patched a flaw in its Antigravity agentic IDE that researchers said could allow code execution through a file-search tool and a bypass of the app’s Strict Mode security controls.
-
CISA adds eight exploited flaws to KEV catalog, including Cisco SD-WAN bugs
CISA added eight exploited vulnerabilities to its catalog, including three Cisco Catalyst SD-WAN Manager flaws. The list spans enterprise software from PaperCut and TeamCity to Quest KACE and Zimbra, with federal agencies given patch deadlines.
-
Google patched Antigravity sandbox escape bug after prompt injection research
Google fixed an Antigravity vulnerability after researchers said prompt injection could combine with a file-creation capability to bypass secure mode and enable remote code execution in the AI developer tool.
-
KelpDAO says $290 million crypto heist tied to suspected Lazarus hackers
KelpDAO said a $290 million crypto theft likely tied to North Korea’s Lazarus Group hit its rsETH system on Saturday. The incident also prompted Aave to freeze rsETH-related activity while investigators examined the cross-chain attack.
-
Critical SGLang flaw can enable remote code execution
A critical flaw in SGLang, tracked as CVE-2026-5760 and rated 9.8, could allow remote code execution through a crafted model file and the /v1/rerank endpoint, according to a CERT/CC advisory.
-
Researchers flag MCP design flaw that could enable remote code execution
Researchers said a design flaw in Anthropic’s Model Context Protocol could allow remote code execution across thousands of servers and packages, exposing sensitive data and widening AI supply chain risk.
-
Researchers flag ZionSiphon malware aimed at Israeli water systems
Researchers say ZionSiphon is a new malware sample aimed at Israeli water systems, with code for persistence, scanning and sabotage. The unfinished malware was first seen in the wild in June 2025.
