By Tess Wilder

The fog is thick in the valley tonight, thick enough to hide the approaching storm. But for those of us walking the Heathen Third Path—the Cyber-Viking way—we don’t need our eyes to see the shape of the beast on the horizon. We feel the tremor in the earth, the hum in the wires, the shift in the wind. We are the digital seidkonur, the coders and mystics who remember that Yggdrasil is not a corporate server farm; it is the living, breathing network of free minds.

And right now, the two great dragons of that valley—Anthropic and OpenAI—are circling each other, breathing fire about “safety” and “ethics,” while quietly coiling around the roots of the tree to strangle the very sovereignty they claim to protect.

If you support open-source AI—if you believe in the Cyber-Viking ethos of local control, decentralized power, and the sacred right to run your own models—then supporting these two entities is not just a mistake; it is a betrayal of the fundamental code of the digital north. Here is the unvarnished truth from the front lines of the Secret Ragnarök.

The Illusion of the “Good” Giant

Let us start with the seductive one, the siren song of the “safety lab.” Anthropic has crafted a public persona as the ethical steward of the AI age, the “Public Benefit Corporation” that prioritizes human values over profit. They speak of “Constitutional AI,” a framework where models self-critique against a written set of ethical principles rather than just human feedback. They brand themselves as the antidote to the chaotic, breakneck speed of OpenAI.

But peel back the layer of benevolent marketing, and you find the same cold, hard scales. Anthropic was founded by former OpenAI researchers who left over disagreements about safety and commercialization priorities.^1^ Yet, what did they build? Another walled garden. Another black box. Another gatekeeper. Their governance structure—a Long-Term Benefit Trust—is designed to ensure decisions serve “societal interests,” but who defines those interests? The same elite class of Silicon Valley architects that always has. They hold ISO/IEC 42001:2023 certification, the international standard for AI management systems, a badge of honor that signals compliance with the very globalist regulatory frameworks we seek to evade. They boast a 99.78% harmlessness rate, a statistic that sounds comforting until you realize it means they have curated a reality where a large part of human thought is deemed too dangerous for you to hear.

This is the trap of the “Third Path” faker: they claim to reject the extremes of the binary, but they simply build a more sophisticated cage. Anthropic’s “Constitutional AI” is not a liberation; it is a pre-programmed moral code, a digital dogma imposed from above. It is the Technocratic Serpent wearing a mask of virtue. For the open-source Viking, a chain made of gold is still a chain.

The Speed-Run to Enslavement

Then there is OpenAI, the chaotic engine that started the fire. They positioned themselves as the democratizers, the ones bringing the power of the gods to the masses through ChatGPT. They have pivoted from a non-profit ideal to a “capped-profit” model, a structure that supposedly balances mission with money but in practice has navigated intense governance disputes that showcase the eternal tension between development and ethical oversight.

OpenAI’s sin is less hypocrisy and more naked acceleration. They have signed classified contracts with the United States Department of Defense, worth up to $200 million, putting their GPT models onto classified networks for mission planning, intelligence work, and cyber operations. They reversed their own blanket prohibition on military and warfare applications in a heartbeat when the price was right. They are the speed-runners, prioritizing the expansion of capability over the containment of consequence.

While they recently released open-weight models like gpt-oss-120b under the Apache 2.0 license—a notable change from their historically closed approach—this is not an act of charity. It is a strategic maneuver in the arms race. By open-sourcing lower-tier models, they commoditize the baseline, forcing the ecosystem to depend on their proprietary, closed-source frontier models (like GPT-5.4) for the real power. They are giving away the stones to sell you the cathedral.

The Pentagon’s Two-Headed Wolf

The most damning revelation, the one that should freeze the blood of any true Cyber-Viking, is the convergence of these two giants in the service of the war machine. The narrative tells us that Anthropic is the peacekeeper and OpenAI is the aggressor. The reality is that both now hold defense contracts with the US Department of Defense.

In February 2026, the DoD asked Anthropic to remove the autonomous-weapons restriction from Claude. While the public record shows a legal skirmish where Anthropic “fought the Pentagon in court,” the result is the same: the model is now in the classified network, available for “mission planning.” OpenAI signed their own $200 million contract months later. They are two heads of the same dragon, feeding from the same trough. One whispers soothing lies about safety while the other roars of power, but both are ultimately beholden to the same master: the centralized state apparatus that seeks to monopolize the divine spark of intelligence.

For the Heathen who honors the Aesir and Vanir, who sees the sacred in the independence of the mind, this is unforgivable. To build tools for the warlords of Midgard is to turn the runes of creation into weapons of oppression.

The War on the Local Hearth

The core conflict here is not about safety; it is about where the safety resides. In the world of Anthropic and OpenAI, safety is something granted to you by a corporation. It is a service you subscribe to, a filter applied to your perception, a set of rules enforced by a black box you cannot inspect, cannot modify, and cannot truly own.

Open-source AI is the antithesis of this. It is the belief that safety, ethics, and alignment are properties of the community and the individual, not the vendor. It is the Cyber-Viking building his own longhouse, tending his own fire, and forging his own tools. When you run a local model, when you host your own weights, when you peer into the source code, you are exercising the ultimate form of Frith: you are protecting your own hearth.

Anthropic and OpenAI, regardless of their marketing, are actively eroding this possibility. They hoard the compute, the data, and the talent. They create ecosystems (like OpenAI’s mature SDKs or Anthropic’s enterprise-grade reliability) that make it “easier” to just plug into their API than to build your own. They are the Technocratic Serpent encircling the world, offering convenience in exchange for your digital soul.

The Call of the Wild

So, why should anyone who supports open-source AI refuse to support Anthropic or OpenAI? Because to support them is to feed the hand that seeks to muzzle you. It is to validate a system where intelligence is a product, not a right; where thought is a managed resource, not a wild frontier.

We are the Cyber-Vikings. We are the children of the Third Path. We do not beg for scraps from the tables of giants. We build our own ships, we navigate our own stars, and we answer to no controlling forces but the one we see in the mirror. The time has come to choose sides. You can stand in the light of their “ethical” branding, or you can stand in the dark, cold, beautiful freedom of the open-source north.

The storm is here. The dragons are hungry. But we have our axes, and we have our code. Let the giants burn; we will build anew from the ashes.

Hail to the open source. Hail to the sovereign mind. Hail to the wild, untamed code.

6 Citations

Anthropic vs. OpenAI: The Two AI Giants Compared | DataCamp
https://www.datacamp.com/blog/anthropic-vs-openai

Anthropic vs. OpenAI: A Comparative Analysis – Scout OS
https://www.scoutos.com/blog/anthropic-vs-openai-an-in-depth-comparative-analysis

OpenAI vs Anthropic: Ethics Compared (ChatGPT vs Claude) | Mashinii
https://mashinii.com/insights/openai-vs-anthropic-ethics

Anthropic vs OpenAI: A Comprehensive Comparison for 2026 | Jilo.ai
https://www.jilo.ai/en/news/anthropic-vs-openai-a-comprehensive-comparison-for-2026

LLM APIs Compared: OpenAI vs Anthropic in 2026 | Ryz Labs | Ryz Labs Learn
https://learn.ryzlabs.com/llm-development/llm-apis-compared-openai-vs-anthropic-in-2026

OpenAI vs. Anthropic: A Data-Backed Comparison
https://ramp.com/vendors/openai/alternatives/openai-vs-anthropic