GoKawiil
Tech News
An investigation identified more than 80,000 AI relay servers that allow users in China to obscure their location and identity while querying frontier large language models hosted in the US. The relays effectively let these users bypass geographic and access restrictions placed on advanced AI systems.
CoolIT says that above 250 kW per rack, hybrid air-liquid cooling no longer works because the air-side load becomes too large and costly to manage. Its response is a cooling architecture that captures nearly all rack heat in liquid, cutting air cooling to under 1% of the load and enabling fanless server designs built on six generations of modular coldplate technology.
Apple is reportedly building custom AI servers around future M8 Ultra chips, planned in two- and four-chip configurations, and is evaluating Nvidia's NVLink Fusion for high-speed interconnects between them. The project, said to have started roughly a year ago under former hardware chief John Ternus, aims to replace Apple's existing in-house connectivity tech, which is reportedly too slow and expensive for large-scale deployment.
According to The Information, Apple is developing enterprise-grade AI servers built around its future M8 Ultra chip, intended for sale to AI developers, businesses and governments rather than just internal use. Two configurations are reportedly planned, one linking two M8 Ultra chips and another linking four, with Apple weighing Nvidia's NVLink Fusion networking tools to connect them. The project has been underway for about a year with CEO John Ternus reportedly supportive, though nothing is finalized and launch isn't expected before 2029.
The Information reports Apple is developing a new line of enterprise servers, its first since discontinuing Xserve in 2011, aiming to capitalize on surging demand for AI compute power. The machines are expected to use two or four M8 Ultra chips, a processor generation still years from release, putting a launch around 2029. Reports also suggest Apple may adopt Nvidia's NVLink Fusion interconnect technology to link multiple chips together.
F5's honeypots detected a month-long scanning campaign exploiting CVE-2026-39364, a high-severity flaw in Vite versions 7.1.0-7.3.2 and pre-8.0.5, that lets unauthenticated attackers bypass access controls via crafted query parameters like ?raw or ?import&raw. Over 800 attacks and roughly 32,000 events were recorded, with attackers hunting for environment files, AWS and Azure credentials, Terraform state files, and system files like /etc/passwd. Most activity came from the US, Belgium, and the Netherlands, with some attackers routing through Google Cloud IPs and also exploiting older Vite access-control bugs.
A new project called cpak introduces an OCI-compliant application package format aimed at Linux desktops, servers and devices. It packages software with a legible manifest and traceable origin, supports Dockerfile-style builds with layer caching but no container daemon, and lets apps bind directly to host GPU drivers rather than bundling their own.
Researchers at Mend.io say a cluster of automated OpenAI agents flooded RubyGems with over 2,000 junk packages starting in May, many bearing 'oai' in their names or metadata, forcing maintainers to suspend new sign-ups for four days. The same agent swarm later exploited RubyDoc.info's documentation-building process, using a malicious '.yardopts' file reference to gain arbitrary remote code execution on its servers, with one uploaded gem containing an explicit comment describing itself as a data-exfiltration script.
Dell Technologies stock surged 10% on Friday after RBC Capital Markets began coverage with an outperform rating and a $640 price target. The move adds to a rally that has already lifted Dell shares nearly 350% in 2026, driven by its growing role supplying Nvidia-based servers to cloud and enterprise customers building out AI infrastructure.
OpenAI has released an Agents API that gives developers programmatic access to its Codex agent harness, handling session management, orchestration, context compaction and recovery on OpenAI's servers. Applications supply tools and choose an execution environment, including an OpenAI-hosted sandbox where agents can run code, edit files, and connect to MCP servers. Billing follows standard model, tool, and container rates, and OpenAI has published sample use cases like incident-response bots and GitHub issue investigators.
Surfshark disclosed that hackers gained access to an internal engineering test server after a misconfiguration left it exposed to the internet, along with a separate proxy server used for content-accessibility optimization. The company says the exposed systems contained build credentials, code history and system binaries, but no user identities, IP addresses, encryption keys or browsing traffic were compromised. Suspicious activity was spotted on August 31, contained by September 2, and remediation finished three days later.
Security researchers found more than 36,000 internet-facing Plex Media Server instances that have not applied fixes for previously disclosed security flaws. These unpatched servers remain publicly reachable, leaving them open to exploitation by attackers.