Cloudflare One gets new capabilities to secure Shadow AI
Cloudflare has added new capabilities to its zero trust platform Cloudflare One to help organizations securely adopt, build, and deploy generative AI applications.
Its new Shadow AI Report provides visibility and analytics of AI activity in general purpose models like ChatGPT, code-generation assistants like GitHub Copilot, and other specialized tools for areas like marketing, data analysis, or content creation.
Once shadow AI use is uncovered, customers can use Cloudflare Gateway to enforce AI policies. They can fully block specific AI applications, limit the types of data uploaded to them, and complete reviews of AI tools to ensure they meet security standards.
Groundcover reveals observability solution for LLMs and agents
The eBPF-based observability provider groundcover announced an observability solution specifically for monitoring LLMs and agents.
It captures every interaction with LLM providers like OpenAI and Anthropic, including prompts, completions, latency, token usage, errors, and reasoning paths.
According to groundcover, while LLMs can offer many benefits, they also introduce a lot of negatives: performance volatility, unpredictable ROI, quality drift, and introduction of security and compliance risks.
Grafana launches preview of AI assistant that works across all of Grafana Cloud
Grafana Labs announced a public preview of Grafana Assistant, an AI assistant that IT teams can use to interact with logs, metrics, and traces in a conversational manner.
It is available in all parts of Grafana Cloud, and sees the context of what is on the page so that it can give specific, context-aware answers. Users can ask questions like “what am I supposed to do on this page?” or “can you explain the data I’m seeing here?”
Grafana Assistant can also guide users through interactive tutorials for completing tasks, such as adding an alert to a metric, adding a new on-call schedule in Grafana Cloud IRM, or how to declare a test incident.
LightBeam.ai provides monitoring for sensitive data usage in Microsoft Copilot with latest updates
LightBeam.ai has announced its Summer 2025 product release, which adds new features to help enterprises secure Microsoft Copilot usage.
Companies will be able to monitor Copilot prompts, responses, and file access to ensure that sensitive data is being properly safeguarded. It will be able to access all documents that are referenced, including Excel workbooks, CSV files, source code, PowerPoint decks, or contracts.
The platform will tag sensitive attributes, check entitlements, and route alerts into the company’s existing governance and privacy workflows.
Bugcrowd launches new products for speeding up vulnerability remediation
Bugcrowd has announced two new products for helping IT and security teams remediate vulnerabilities more quickly.
The first, AI Connect, is like a read-only “front door” for vulnerability data, Bugcrowd said. It allows customers to integrate Bugcrowd vulnerability data into their preferred internal AI stacks. Under the hood, it is built upon MCP to enable the connections.
“As security teams adopt AI-driven tools to automate workflows and scale operations, they often hit a critical roadblock: their AI systems lack access to real-time, high-value security data. Without direct integration into bug bounty, penetration test, and red team findings, teams are forced to manually cross-reference reports with internal systems. This slows response times and reduces the accuracy of remediation, creating a widening gap between AI’s potential and its practical impact on security outcomes,” Bugcrowd wrote in an announcement.
