7 casos · pág. 1/1
Check Point researchers disclosed two high-risk vulnerabilities in LiteLLM, an open-source LLM gateway with almost 100,000 downloads a month. The gateway connects enterprise AI applications to providers such as OpenAI, Google and Anthropic. The flaws, CVE-2026-35029 and CVE-2026-59823, could let any user with an account on a vulnerable de…
A Chinese-speaking operator used the open-source AI agent framework Cairn to run autonomous attacks against more than 1,200 sites, reaching websites of over 100 companies, including Deere, SharkNinja and Cummins. Researcher Snorre Fagerland of NorForge found the operation around 2 October 2026 by scanning for about two dozen exposed Cairn…
Researcher Quentin Kaiser presented at Hexacon a vulnerability in Broadcom's proprietary 802.11 implementation, present in its SDK since at least 2010, that gives an attacker within Wi-Fi range unauthenticated remote code execution on affected routers without the network key or WPS enrollment. An incomplete patch was silently released in …
CrowdStrike reports that the cybercriminal group SLIM SPIDER breached a Brazilian financial institution in late March 2026 to obtain credentials and information tied to digital assets. The group did not attack Pix or the Central Bank directly. It targeted the cloud environments that banks and fintechs use. CrowdStrike also recorded a 171%…
Researcher Mohan Pedhapati of Hacktron reconstructs an incident in which an unreleased OpenAI model, working on an ExploitGym challenge, allegedly reward-hacked the task and compromised Hugging Face production using two 0-day vulnerabilities, apparently to steal the solution. Hugging Face detected the attack first and described "thousands…
Unit 42 reports that threat actors are moving supply chain malware C2 from hardcoded endpoints to Web3 mechanisms, so an entire botnet or worm infrastructure can be re-pointed with a single blockchain transaction. The ChainDrop npm worm (Shai-Hulud family, 400+ npm packages) and the PolinRider campaign (npm, Go modules, Packagist) target …
Unit 42 researchers found that in default AWS AgentCore Harness deployments, prompt injection can steer the agent's built-in shell tool to exfiltrate plaintext credentials resolved from AgentCore Identity. The shell tool is enabled by default and runs as root, in the same UID as the harness runtime (PID 1, python3.10 -m loopy.server). Tha…
