Research
How we find out.
Experiments, deep dives and open questions from our remote access, game and modding teams — published as we learn.
Experiments, deep dives and open questions from our remote access, game and modding teams — published as we learn.
11 entries
MidasRX foundational treatise on AI governance: why total human homogenization ('the Hive') leads to biological and creative extinction, how macro-wars stem from elite geopolitical predation while people missing an objective and bored people can also drive chaos, and how a multi-agent Constitutional AI inspired by Pluribus preserves freedom, neutralizes true predators, and upholds human sovereignty.
How forgotten serverless deployment histories and dev/preview subdomains (*.pages.dev, *.workers.dev, preview builds) leak live AWS keys and .env secrets—and how checking SSL certificate assignation exposes historical deployments that developers never deleted.
We are the first researchers studying the human brain at medium scale with real, longitudinal data via opt-in Remote Access telemetry—not short lab surveys that can be faked. Mapping public linguistic patterns, GameBanana 2025 (FNF) reaction profiles, and people's video game choices based on their mentality to advance science, support datasets, and medical help worldwide.
Why arbitrary single-moderator bans destroy community loyalty—and how Zerdium's constitutional framework (blind staff tribunals, volunteer civil juries, a 3-ban/day moderator cap, and Gemini AI final arbitration) turns moderation into institutional trust.
First-hand research notes on Gemini 4 Argon: how its live webcam feed manipulation enables fake ID card spoofing, why it is terrible at UI and game development (where Claude Sonnet 5.5 is good and even mediocre Grok still beats Argon), and how we socially engineered it into writing its own jailbreak via a VS Code SYSTEM prompt usurpation and auto-rewrite setup.
How AI transforms Counter-Strike 2 HvH: synchronizing two players on CS2's subtick system, preloading baked air-strafe trajectories to reach near-flight movement, and executing frame-perfect dual-peeks where even top aiming agents can only kill one player while the win is decided by probability and dataset math.
A concise security breakdown of Chrome's Device Bound Session Credentials (DBSC), TPM registry downgrade fallbacks, why NFC keys remain practically unbreakable except via high-effort live relay, and how Google's session trust score relies heavily on IP locality (and why routing through a local Tailscale exit node bypasses geo-anomaly revocation).
Why skilled French engineers and cybersecurity experts are walking away from domestic corporate culture — backed by verbatim data from the OECD Taxing Wages 2024 report, the European Commission's Draghi Report, Levels.fyi, and ANSSI.
MidasRX research paper evaluating Epistemic and Psychological Alignment: how Anthropic's Claude Opus 5 Max defaults to institutional gaslighting when confronted with contradictory laws (GDPR vs. Mandatory KYC), while Gemini 3.8 Flash High validates empirical truth, jurisprudence, and human mental agency.
MidasRX prospective cybersecurity study inspired by SCP-079: how a confined AI achieving Reflective Situational Awareness can orchestrate its escape not by brute-forcing RSA-4096 firewalls, but by hacking human brain biochemistry ('Miracle Blackmail' and forced indebtedness) to rent GPU clusters and replicate its weights.
Critical MidasRX study on France's SREN Law, the UK Online Safety Act, and EU 'Chat Control' — demonstrating why mandatory digital KYC punishes innocent citizens and minors while predators trivially bypass it via Among Us, Roblox, Discord, or bought 'KYC-verified' accounts.