Slack AI evolved over four phases from AWS SageMaker to Amazon Bedrock (provisioned then on-demand) and finally GCP Vertex AI, building an intelligent multi-cloud routing layer with circuit breakers, A/B testing, and hybrid provisioning to ensure resilience, model-best-fit, and zero provider lock-in.

Trail of Bits bypassed malicious skill scanners from ClawHub, Cisco, and skills.sh using simple techniques: newline padding, .pyc bytecode poisoning, .docx indirection, and prompt injection. Existing scanners are structurally broken; organizations should use curated, trusted skill sources instead.

Any web page a victim asks ChatGPT to summarize can become a phishing payload. P0 Labs research reveals a Markdown rendering vulnerability in ChatGPT's response UI.

Post which explores, names, defines, and threat-models an Azure identity type that has quietly operated inside every customer tenant. Never documented under a single name, never owned by you, and never fully visible to you.

Multiple official @redhat-cloud-services npm packages were compromised with a credential-stealing worm derived from the open-sourced Mini Shai-Hulud malware, targeting cloud credentials, and developer tooling across CI/CD pipelines.

DBSC (Device Bound Session Credentials) is now GA and on by default in Chrome for Windows for all Google Workspace users. It binds session cookies to the authenticated device, mitigating session theft even if malware is present.

Claude Code's /security-review is vulnerable to model anchoring bias when run in the same session that wrote the code. A new diff-scoped plugin avoids this but misses cross-commit vulnerability chains where each individual change appears benign in isolation.

A VSCode bug allows stealing GitHub OAuth tokens via a malicious repo opened in github.dev. JavaScript in a Jupyter notebook webview abuses postMessage keydown forwarding to install a local workspace extension, which exfiltrates the token granting full repo access.

Open Source realtime backend in 1 file.

Lightweight server monitoring with historical data, docker stats, and alerts.

Run a fully self-contained, self-improving AI agent from a single folder or USB drive.

Policy enforcement, zero-trust identity, execution sandboxing, and reliability engineering for autonomous AI agents.

Jun 22-26, 2026 | Vienna, Austria