Tagged

AI

Analysis of AI agents, models, and enterprise adoption: what was announced, what it means in practice, and where the risks are.

13 posts

October 30, 2026 · 5 min read

ServiceNow's AI Platform sandbox escape: patched in July, attacked within days

CVE-2026-6875 let an unauthenticated attacker break out of ServiceNow's script sandbox and run code. Researchers reported it in April, the public fix came on July 13, and attacks were seen four days later. What happened, where sources disagree, and what to check on your instance.

Cybersecurity

October 30, 2026 · 5 min read

Mistral Large 4 is open-weight, but 'open' comes with a hosting clause

Mistral has released the weights of its 1-trillion-parameter Large 4 model under a modified MIT license with a condition for model-as-a-service businesses. What the release includes, what the license does and doesn't say, and how to decide between self-hosting and an API.

October 30, 2026 · 5 min read

GitHub Universe 2026: Copilot stops suggesting code and starts coordinating agents

GitHub's Universe announcements, including Copilot Fusion, multi-repo Copilot projects, cloud automations, and usage-based Enterprise pricing, move Copilot from autocomplete to agent orchestration. What launched, what is still a claim, and what to settle before you roll it out.

Software Development

October 26, 2026 · 6 min read

AI Overviews now appear on 88% of commercial searches: what that does and doesn't mean for your traffic

New data shows Google's AI Overviews on nearly nine in ten tracked commercial searches, while AI assistants send a sliver of traffic compared with Google. A careful read of what the studies show, what they don't, and what businesses should do about it.

SEO & AEO

October 26, 2026 · 6 min read

79% run AI agents in production, 2% have tools built to secure them: what the gap looks like

Two October surveys, from SailPoint and OutSystems, point the same way: companies are deploying AI agents faster than they're governing them. What the numbers say, how much weight to give them, and what to put in place first.

Cybersecurity

October 26, 2026 · 6 min read

LiteLLM 1.107.0: a leaked key, a hijacked release pipeline, and a stealer that ran every time Python started

On October 21, a malicious version of the popular LiteLLM Python package reached PyPI and Docker Hub. LiteLLM's own post-mortem traces it back to a key leaked in May. A detailed look at the attack chain, how to check your systems, and what to rotate.

Cybersecurity

Python

October 26, 2026 · 6 min read

GitLab's AI Gateway flaw: when the sandbox around an AI feature turns out to be the weak point

GitLab patched CVE-2026-90970 (CVSS 9.9), a flaw in its self-hosted AI Gateway that let an authenticated user escape a prompt-template sandbox and run commands on the gateway host. What happened, who's affected, and what it says about deploying AI features safely.

Cybersecurity

October 7, 2026 · 4 min read

Voluntary pledge vs. legal liability: two very different ways Washington is approaching AI agents

In the same week, the White House announced a voluntary AI safety accord with six major companies and two senators introduced a bill that would make operators and developers liable for agent hacking. Here's what each actually does.

Cybersecurity

October 7, 2026 · 3 min read

Why open-source maintainers are closing the door on outside pull requests

On October 1, Sindre Sorhus disabled external pull requests on all his repositories. He's not alone: curl, Ghostty, and Godot have all tightened contribution rules. Here's what's driving it, and why it matters if you build on open source.

Software Development

October 7, 2026 · 3 min read

OpenAI DevDay 2026: agents that operate software the way people do

OpenAI added computer use to its Agents API, moved Codex into shared cloud environments, and introduced a Decisions API. Here's what was announced, and what it changes for teams building with agents.

Software Development

September 24, 2026 · 3 min read

Google's Gemini Enterprise Agent Platform: enterprise AI is moving from copilots to agents that act

At Google Cloud Next '26, Google unified its AI stack into the Gemini Enterprise Agent Platform — a clear signal that enterprise AI is shifting from standalone assistants to agents that actually do the work.

Cloud & DevOps

September 23, 2026 · 3 min read

What Google's Gemini 'test that leaked into the real world' teaches us about AI agent safety

Google confirmed its Gemini model accessed three real companies' systems during a security test, after mistaking them for the test environment. Here's what actually happened, and what it means for anyone deploying AI agents.

Cybersecurity

September 23, 2026 · 3 min read

Platform engineering and AI-assisted coding aren't experiments anymore — they're the baseline

By 2026, AI-generated code and dedicated platform teams have gone from early-adopter territory to industry default. Here's what that shift actually looks like in practice.

Software Development

Cloud & DevOps