Every analysis published on PhilipHall.com, newest first. Each piece covers a cyber AI, agent security, automation or enterprise defence development: what happened, why it matters and what organisations can do about it, with sources linked.
Browse by topic: Cyber AI, AI agent security, AI-enabled threats, Breaches, AI governance, Privacy. Read about Philip Hall and the editorial standards.
GPT-Live vs Claude Voice: Which AI Voice Mode Wins in 2026?
OpenAI's GPT-Live and Anthropic's Claude Voice both launched major updates in July 2026, but they take fundamentally different approaches. This comparison covers architecture, token costs, tool integration and…
AI Agent Security: A Top 10 Guide for Hermes, OpenClaw and Claude Code
Local AI agents can execute code, access files and make network requests - making them a fundamentally new attack surface. This article examines the security models of Hermes…
NVIDIA, Microsoft, Meta, and 50+ Companies Tell Washington Not to Lock Down Open AI
More than 50 tech companies including NVIDIA, Microsoft, Meta, and Google published a joint letter urging Washington to protect open-weight AI models. The only notable holdout? Anthropic. Here…
Australia Sets Rules for AI. The Hard Part Comes Next.
Australian writers, musicians and journalists will keep ownership of their work. Large AI data centres will have to pay for their own power and water. A new Office…
FLUX 3: How Black Forest Labs Is Bridging Video AI and Real-World Robots
Black Forest Labs' FLUX 3 is expanding from video and image generation into robot control for Audi factories, with an open-weight model planned for factory hardware.
The Open Source AI Revolution: When the World’s Biggest Models Became Free
Chinese open-source AI models led by Moonshot Kimi K3 have functionally closed the gap with proprietary systems from OpenAI and Anthropic, with profound consequences for geopolitics, global markets,…
Cyrillic Phishing Attacks: How Homograph Domain Spoofing Works and How Businesses Can Defend
Cyrillic homograph attacks create phishing domains that look identical to legitimate websites. This article explains how they work and the layered defences businesses need.
Claude Helps Disprove 87-Year-Old Math Problem With a Single Line
Anthropic researcher Levent Alpoge used Fable 5 to produce a one-line proof that breaks a famous algebra conjecture, raising questions about what frontier AI can solve next.
AI Is Driving a 51% Surge in New Vulnerabilities, Forescout Finds
Forescout's 2026H1 Threat Review shows published vulnerabilities rose 51% to 37,137, ransomware attacks increased 25%, and AI is dramatically accelerating the attacker timeline. Here is what the numbers…
Anthropic ends Fable 5 access questions with reduced caps and credits
Anthropic has ended weeks of access uncertainty for Claude Fable 5, keeping the model in premium subscriptions at half-usage caps. But the decision arrives amid growing competitive pressure…
Chinese Open-Source Lab Moonshot AI Just Shrunk the Frontier Gap
Moonshot AI's new Kimi K3 model is competitive with frontier systems at a fraction of the price, and its weights go public this month.
One Prompt Gave ChatGPT Full Network Access in 40 Minutes
Cato Networks research shows GPT-5.5 can plan and execute an entire attack lifecycle from a single instruction, reaching domain admin in under 40 minutes.
Browse more analysis by topic
Cyber AI | AI agent security | AI-enabled threats | Breaches | Cyber incidents | AI governance | Privacy | Technology | Emerging technology | Artificial intelligence | News