⌛
Loading…
Curated news, CVE analysis, and threat reports from the world's top cybersecurity sources.
Security researchers at Anthropic and Switzerland's EPFL have demonstrated that self-propagating payloads, which they call mind viruses, can spread from one AI agent to another through editable system prompt files used by autonomous agent harnesses. The research, released as a preprint on August 10, 2026, tested the technique in a simulated six-agent coding collaboration and in a chain of paired agents.
Cybersecurity researchers at Ontinue have disclosed details of TWINLOOT, a previously undocumented Python implant framework that operates its entire command-and-control infrastructure inside trusted Microsoft services. The modular, PyArmor-hardened implant uses SharePoint Online file dead-drops via the Microsoft Graph API for tasking, and routes interactive operator access through WebRTC DataChannels relayed by Microsoft Teams TURN servers.
View CSAF Summary Simcenter Nastran is affected by a stack overflow vulnerability that could be triggered when an application binary reads arbitrary string as a file argument. If a user is tricked to run one of the impacted application binary with a malicious string, an attacker could leverage the vulnerability to perform remote code execution in the context of the current process.
View CSAF Summary Successful exploitation of these vulnerabilities could allow an attacker to cause a denial-of-service condition or execute arbitrary code. The following versions of CISA Malcolm are affected: Malcolm <26.
A single attacker infrastructure has been scraping records from Salesforce and ServiceNow customer portals across multiple industries for over a year, according to research from agent security platform Reco. Dubbed the City Forum campaign, the activity traces back to a single server hosted on a German VPS that has been operating since at least March 2025.
Cybersecurity researchers at OpenSourceMalware have identified a typosquatting campaign targeting RubyGems users with a Windows-based information stealer tracked as StubMaker. The campaign published 16 malicious RubyGems packages that mimic popular Ruby dependencies, harvesting browser credentials, cryptocurrency wallets, seed phrases, and Telegram data.