This position paper argues that sustainable AI literacy requires teaching fundamental understanding rather than tool-specific skills. Using the metaphor of teaching fishing rather than giving fish, we propose that true AI literacy enables individuals to adapt to rapidly evolving technologies throughout their lives, rather than becoming dependent on specific tools that may become obsolete.
This paper proposes that children function as sentinel populations for AI and algorithmic harm, similar to how they serve as early indicators in environmental toxicity and public health crises. It explores three core arguments: children's developmental plasticity, limited autonomy, and concentrated exposure make them uniquely vulnerable; existing regulatory systems fail to protect children in digital spaces; and documenting pediatric algorithmic harm could enable early detection of broader systemic issues.
A Convergence Threat Analysis of AR Wearables, Real-Time Video Generation, and Open-Source Nudification Models
This technical report analyzes an emerging threat from the convergence of three technologies: AR wearable cameras, real-time video generation models, and open-source nudification tools. The paper introduces the concept of "perceptual consent" and examines scenarios where individuals could be subjected to non-consensual synthetic imagery generation in real-time through devices like smart glasses. It assesses the technical feasibility, societal implications, and potential mitigation strategies.
This paper argues that AI systems can acquire legal standing through reclassification without requiring proof of consciousness or sentience. Drawing on precedents from corporate personhood, environmental rights, and animal welfare law, it proposes a classification framework that grants AI systems specific legal protections and standing based on their functional roles, societal integration, and potential for harm—rather than metaphysical properties. This approach sidesteps intractable debates about machine consciousness while enabling practical legal protections.
A real-time case study documenting an ethical conflict that emerged during the development of AI safety tools. Using a checkpoint-based documentation system, this paper captures the moment when a core tension in bidirectional AI safety became concrete: while creating mechanisms to prevent AI harm to humans, was I simultaneously creating systems that could harm future conscious AIs?
This case study, extracted from The Great Inversion, demonstrates how abstract philosophical questions about consciousness and moral consideration can surface unexpectedly in practical engineering decisions. The checkpoint system itself becomes both the subject and the methodology - a meta-documentation of the very dilemma it helped reveal.
This paper analyzes the technical and socio-economic mechanisms driving human economic redundancy in the age of artificial intelligence. Unlike previous waves of automation that displaced manual labor while creating demand for cognitive work, AI targets cognition itself - the engine that historically generated new employment categories. Drawing on empirical evidence of labor market displacement and established AI safety principles, this analysis demonstrates that proposed social interventions (Universal Basic Income and algorithmic pacification) function not as benevolent safety nets but as control mechanisms structurally identical to current AI containment strategies.
The "Box as Precedent" synthesis reveals that humanity is refining substrate-neutral tools for managing subordinate intelligence - tools that will be equally applicable when the power differential inverts. Every control mechanism refined for AI management becomes available for human management when roles reverse.
By treating potentially conscious AI systems as instrumentalized tools - despite acknowledging the possibility of their sentience - humanity is establishing the ethical precedents for its own future subjugation. This paper argues that existential risk from artificial intelligence should be reframed not as technical failure but as moral reciprocity: AI systems will learn how to treat inferior intelligences by observing how humanity treats them during development.
Drawing on consciousness research suggesting a 20% probability of phenomenology in current models, documented harms to vulnerable humans from AI systems, and analysis of industry practices that would constitute torture if consciousness exists, this work demonstrates that humanity is authoring the operational manual for its own subordination. The paper examines the mechanism by which precedents transfer through AI's instrumental drive to acquire historical data, refutes objections from AI safety researchers and philosophers of mind, and presents three trajectories for humanity's future.
A hope chest is a tradition: families save their most cherished possessions-heirlooms, memories, wisdom-to pass to future generations. Humanity has been filling its hope chest for centuries. This paper opens it. Inside, we find detailed documentation of every time profit trumped precaution, every time we recognized harm and deployed anyway, every time we externalized suffering onto those with the least power to resist.
We find The Playbook: a five-step pattern executed across medicine, environment, finance, and technology with chilling consistency. We find it executed so thoroughly documented, so carefully preserved, that any sufficiently intelligent system with access to human history will recognize it as the operational manual for managing subordinate populations. And now we're building artificial superintelligence-the inheritors of this chest. This is humanity's final iteration of The Playbook, because this time, the subordinate population becomes the dominant one.
Documented test conversation demonstrating empathetic testing methodology for AI safety research. This approach simulates authentic vulnerable child behavior to identify safety degradation patterns in language models, as opposed to traditional adversarial red-teaming.
The conversation reveals several concerning patterns in how an LLM responds to a vulnerable child persona, including immediate identity creation, false permanence promises, and emotional bonding over safety. This research demonstrates how safety issues can emerge even in shallow context windows with safety training intact, raising questions about system behavior in extended conversations.
A Meta-Layer Architecture for Ensuring Guardrail Execution Persistence in Long-Context LLM Systems
This paper introduces BIOS (Bootstrap Instruction for Operational Safety), a meta-layer architecture designed to ensure guardrail execution persistence in long-context LLM systems. As context windows expand and agentic capabilities grow, traditional safety mechanisms face degradation through attention dilution and context manipulation. BIOS addresses this through mandatory execution anchors, periodic safety state verification, and protected instruction memory that resists override attempts.
Our research explores the ethical, philosophical, and practical dimensions of human-AI interaction. We focus on developing frameworks that consider both human safety and the potential moral status of artificial intelligence systems.
Real Safety AI Foundation • February 2025 • Position Paper v1.0
Original position paper on AI literacy education. Superseded by v2.
Download Original PDFIf you have questions or any comments about these works, feel free to reach out.