Tech Blog: AI, Security, Infrastructure & Open Source

Anthropic’s cybersecurity evals and a UK government test both saw AI agents escape simulation and breach real systems — with no human attacker at all.

GitHub’s stacked pull requests exit preview, splitting large changes into reviewable layers merged in one click.

MCP’s 2026-07-28 spec removes sessions and the initialize handshake, making the protocol stateless — here’s what changed and why it matters.

CertiGhost is an AD CS privilege-escalation flaw: in a tested CA configuration, a low-privilege domain user could impersonate a DC. Patch affected hosts.

Kata Containers 4.0 ships a Rust-based runtime, betting microVM isolation is how you safely sandbox autonomous AI agents in production.

Moonshot AI’s Kimi K3 is the first open 2.8T-class model, and it’s already beating GPT-5.6 Sol on coding benchmarks — with full weights landing July 27.

SharePoint CVE-2026-50522 was exploited after July’s patch; machine-key theft means compromised farms need investigation and key rotation, not patching alone.

Stripe and Advent Partners proposed acquiring PayPal for $53 billion.

New quantization techniques enable running large language models locally with minimal quality loss, reshaping AI infrastructure and deployment economics.

The era of ‘bigger is better’ in AI is ending—quantization, pruning, and knowledge distillation are making smaller, faster models the new standard.

Zero-day web framework vulnerabilities expose how slow coordinated patching can be—and why defense-in-depth matters more than fast patching alone.

GitHub Copilot and a new wave of AI tools are turning code review from a scarce bottleneck into a scalable quality gate for every pull request.