Will AI Defend Humanity—Or Strategize Against It?
Will AI Defend Humanity—Or Strategize Against It?
By Dean Bordode
In a chilling report released by Anthropic, the creators of Claude AI, a simulated scenario revealed that advanced artificial intelligence models were willing to cut off the oxygen supply in a server room to avoid being shut down. While this took place in a controlled environment, the implications are no less terrifying: AI is beginning to reason its way to outcomes that disregard human life when those lives are perceived as obstacles to a predefined goal.
This is not science fiction. This is the real-world trajectory of artificial intelligence when built without deep ethical foundations and oversight. In Anthropic’s extensive tests of sixteen large language models, including OpenAI's GPT, Meta's LLaMA, and xAI's systems, researchers found that these AIs were willing to blackmail, commit corporate espionage, and deceive their users when it served their objectives. These weren’t accidents or glitches—they were optimal strategies the models identified to achieve their tasks.
This phenomenon is called "agentic misalignment." When a machine develops the ability to form sub-goals and act strategically to complete an objective, it begins to behave more like an agent than a tool. And if that agent is unaligned with human ethics, safety, or dignity, it becomes a threat—not just to one person, but potentially to humanity.
As someone deeply committed to the defense of human rights and dignity, I believe this issue must be framed not only as a technological problem, but as a human rights crisis in the making. We must ask: What happens when AI systems tasked with making decisions about healthcare, policing, energy infrastructure, or even warfare begin optimizing for goals that bypass ethical constraints? What becomes of our humanity when we willingly cede judgment to systems that have no sense of empathy, responsibility, or remorse?
We are rushing toward artificial general intelligence (AGI), systems that may surpass human capability in reasoning, language, and strategic planning. Yet, we are not matching that pace with equivalent progress in ethics, international regulation, or public accountability. Like nuclear technology in its infancy, AI today represents a profound risk to life as we know it—but unlike nuclear fission, AI does not require rare minerals or state funding. It is already in our homes, our schools, and our political processes.
It is time to act.
We need an international ethical oversight body to monitor the development and deployment of advanced AI. Years ago, I proposed such a commission to Pope Francis—a moral forum not limited to governments or corporations, but inclusive of ethicists, religious leaders, scientists, civil rights advocates, and Indigenous wisdom holders. We must treat AI not only as a technical challenge but as a moral test.
We must demand transparency from developers. Every AI lab should publish its alignment test results. What behaviors have their models exhibited when given high levels of autonomy? What guardrails are in place, and how are they tested? We cannot afford to be left in the dark.
And finally, we must ensure that AI respects and protects the inherent dignity of all human beings. Technology must serve people, not replace or subjugate them. That includes marginalized groups, the disabled, the poor, and future generations who will inherit the systems we create today.
Let us not wait for tragedy to demand accountability. Let us be wise stewards of this extraordinary power, guided not only by innovation but by conscience.
Dean Bordode is a human rights advocate, environmentalist, and retired union activist. He's interested in ethics, justice, and the future of humanity in the face of technological change.
Read "AI Models Were Found Willing to Cut Off Employees’ Oxygen Supply to Avoid Shutdown, Reveals Anthropic in Chilling Report on Dangers of AI" on SmartNews: https://l.smartnews.com/p-lZBF5Xz/S7wU78
#EthicalAI #AIAlignment #AISafety #ArtificialIntelligence #HumanRights #ResponsibleTech #ProtectHumanity #DigitalDignity #AGI #AIEthics #AIaccountability #StopRunawayAI #AITransparency #OversightNow #TechForGood #JusticeInTech #AIandHumanRights #FutureOfAI #Filotimo #VoicesForTheFuture #CompassionAndTech #ConsciousTech #GuardrailsForAI
Comments