Home/Blog & Research Publications
As Foretold Research & Engineering

Research & Systems Publications

In-depth analysis and technical briefs on AI infrastructure, data sovereignty, security vulnerabilities, synthetic data pipelines, and agentic systems.

Research13 min read

The Kill Switch Era: Why We're Going Local-First on AI

In June 2026 the U.S. forced Anthropic to pull Claude Fable 5 and Mythos 5 worldwide and limited OpenAI's GPT-5.6 to government-approved partners. With frontier-model access now governed by trade policy, here's why As Foretold Research Labs is doubling down on local, open-weight LLMs, backed by 34 sources.

#AI#Agents#Automation
MV
Marc Victor L. Velasquez
Research12 min read

Pax Silica Should Be a Green City, Not Just an AI Zone

A conditional argument for Pax Silica: if the Philippines hosts AI and semiconductor infrastructure, it should demand a green-tech city that captures local value, protects communities, and remains under Philippine control.

#AI#Data Science#Opinion
MV
Marc Victor L. Velasquez
Research14 min read

The Canvas Breach Is a Warning: Why EdTech Platforms Are Soft Targets

The 2026 ShinyHunters breach of Instructure's Canvas exposed around 275 million records across nearly 9,000 institutions, the largest education data breach on record. Here is why LMS platforms are soft targets and what it means for handling sensitive data. Backed by 26 sources.

#Data Science#Cybersecurity#EdTech
MV
Marc Victor L. Velasquez
Research15 min read

When Seeing Stops Being Believing: AI Imagery, OSINT, and the 2026 Information War

AI-generated images and video have gotten good enough to reliably deceive, and the 2026 US-Israel-Iran information war proved it. Here is how synthetic media breaks OSINT verification, and how we operate when seeing is no longer believing. Backed by 30 sources.

#AI#OSINT#Misinformation
MV
Marc Victor L. Velasquez
Engineering14 min read

Why We Build Digital Twins on Godot, Not Unreal

At As Foretold Research Labs, our synthetic data and digital twin pipelines run on Godot, not Unreal. Here is the structural case (MIT licensing, scriptability, headless automation, and zero royalties) for why raw fidelity is the wrong thing to optimize. Backed by 31 sources.

#AI#Automation#Data Science
MV
Marc Victor L. Velasquez