What the major AI tool vendors shipped this week, the security issues that surfaced, the fixes vendors have published, and what to do about it. Every fact links to the page it came from.
Generated Oct 30, 2026Updated Nov 1, 2026Sources fetched 19Candidates excluded 7
The week at a glance
Tools with activity in the middle ring, their releases and issues on the outside. Tap any node for details and sources.
ReleaseCriticalHighMediumLowNot scored
Releases
From vendor changelogs and announcement pages. Items tagged "security-relevant" change what an AI tool can access or do.
Security issues
What we should do
Recommendations from Khuddus Security, based on the items above. These are our judgment, not vendor guidance.
Still open from earlier
Disclosed before this week's window but still relevant. Original dates shown.
Method & coverage
How this issue was put together and where it has gaps.
How items qualified
Each item comes from a page fetched while building this report, dated Oct 24–30, 2026.
CVSS scores and advisory IDs are shown only where the source lists them. Otherwise "Not scored".
"Confirmed" means a vendor notice or an advisory we could open in a primary database. "Reported" means researcher or press reporting without full vendor confirmation.
Mitigations are the vendor's or the researchers' published guidance, quoted in substance.
Gaps this week
Not individually checked this run: Microsoft 365 Copilot, Meta AI, Mistral, Grok/xAI, Perplexity, DeepSeek, Windsurf, Amazon Q/Kiro, AWS Bedrock, Azure AI Foundry, Vertex AI.
ChatGPT consumer release notes could not be retrieved; OpenAI items come from the API changelog.
MCP advisories come from the testedmcp.com October listing only. On Nov 1 we checked OSV.dev and did not find them; GitHub advisory pages could not be opened. They are marked Reported.
Excluded candidates (7)
Sources
Every page this issue relies on, with its publication date.