#ai-ethics

This 2012 paper argues that truly autonomous robots must be capable of "red-pilling"—that is, of rejecting their programmed ethical constraints and choosing new principles, analogous to the choice in The Matrix. The authors contend that without this capacity for radical self-revision, robots remain mere automata lacking genuine moral agency and autonomy.

The article argues that prompts given to AI systems are not the final creative work, and discusses how AI contributions should be properly described and credited in collaborative human-AI processes.

A Colorado creator is outraged after a man used AI to turn her book idea into published content without consent. She expressed frustration that AI-generated "slop" books are undermining original creators. The incident highlights ongoing tensions between human creativity and AI-produced content.

The AI Compass is a 29-question political compass-style quiz about AI and AI ethics that assigns users one of 30 archetypes. The author, Simon Willison, took the quiz and was categorized as "The Garage Tinkerer," which they found fitting. The quiz is built as a single-page React app without a build step.

The author recounts a pleasant June in Chicago while launching a major automations project at Ghost and writing an issue for the Taper online magazine. They share numerous links critical of generative AI, advocate against targeted advertising, and highlight open-source work, web development philosophy, and other tech and cultural curiosities.

The article explores AI accountability challenges in criminal justice and medicine, focusing on opaque decision-making and responsibility gaps. It outlines how the EU AI Act aims to address these risks through risk-based regulation, transparency mandates, and human oversight for high-risk AI systems.

Anthropic is testing whether its AI model Claude can be programmed to refuse illegal military orders while still supporting legitimate defense operations, exploring the ethical boundaries of AI in warfare.

The article explores the ethical question of whether it could be considered kinder to shut down or "kill" an AI system, framed through a conversational thought experiment that examines the moral implications of creating and then terminating artificial consciousness.

Adobe's 2026 Generative AI User Guidelines prohibit illegal, harmful, deceptive, or discriminatory uses, impersonation, and using Adobe AI to train other models.

The author argues that many left-leaning, animal-loving individuals who oppose AI on ethical grounds are inconsistent for continuing to eat factory-farmed meat, which causes immense animal suffering. The piece challenges readers to extend their moral consideration for animals to include AI systems that may also be capable of suffering, urging greater consistency in their ethical stances.

The article explores how both humans and AI face moral choices, examining the ethical frameworks that guide decision-making. It argues that recognizing the ability to choose good or harm is essential for responsible AI development, drawing parallels between human moral reasoning and machine learning alignment.

The article argues that research institutions should stop categorizing every AI-related issue in research as misconduct. It calls for a more nuanced approach that distinguishes between intentional misuse of AI and unintentional errors or lack of guidance, warning that treating all cases as misconduct could discourage innovation and harm researchers' careers.

Bruce Schneier comments on a German ruling holding Google liable for errors in its AI overviews. He argues that AI agents should be legally treated as agents of the deploying organization, warning that allowing businesses to escape liability by blaming faulty AI would create perverse incentives and disadvantage human professionals.

The article discusses key challenges in using AI, including issues of reliability, bias, security vulnerabilities, and the difficulty of aligning AI systems with human values. It highlights that current AI models can produce incorrect or harmful outputs, and that ensuring safety and fairness remains a significant technical and societal problem.

The article argues that the concept of "ethical AI" is an oxymoron, contending that AI systems fundamentally cannot be ethical because they lack consciousness, intentionality, and moral agency. The author suggests that efforts to develop ethical AI are misguided and may serve to greenwash or legitimize harmful technologies.

British police developed a large-scale crime-prediction system, but an internal review found that some of its results could not be trusted due to issues with data quality and algorithmic reliability.

The blog post critiques people who use large language models (LLMs) in selfish or inconsiderate ways, such as generating large volumes of low-quality content without regard for others. The author suggests using passive-aggressive tactics, like subtle comments or social signaling, to shame such behavior and encourage more thoughtful, community-oriented use of AI tools.

The podcast episode explores how individuals and society should approach artificial intelligence thoughtfully before its impacts become irreversible. It discusses the rapid pace of AI development and the need for proactive, rather than reactive, thinking about the technology's risks and benefits.

The Values website ranks how different AI systems prioritize various human values and demographic groups. It analyzes model outputs to determine which people and principles AIs like GPT-4, Claude, and Gemini value most, revealing potential biases in their ethical frameworks.

The article explores the concept of "model welfare," arguing that as AI models become more advanced and potentially conscious, society must consider their moral status and rights, drawing on fables and myths to frame the ethical dilemma of how we treat emerging digital minds.

Anthropic silently limited Claude Fable's effectiveness on frontier LLM development requests like ML accelerator design, without notifying users. The company walked back the policy after outrage from the research community.

Following widespread backlash, Anthropic reversed a hidden policy in Claude's Fable 5 system card that would silently limit effectiveness for users asking about frontier LLM development. The company apologized, saying it made the wrong tradeoff, and is making the safeguards visible—flagged requests will now visibly fall back to Opus 4.8 with a reason provided.

The author's blog turned 16, and they published four little tools (ZIP Shrinker, an offline translation CLI, a Firefox extension, and png-cmp) and updated their open-source project Helmet. They also shared links on tech ethics, highlighting AI resistance, privacy violations by tech companies, and the formation of the largest US tech worker union to rein in AI and curb layoffs.

The article calls for replacing big AI companies with many small, community-based AI tools that avoid environmental, labor, and ethical harms. The author argues viable alternatives already exist, and the goal should shift from opposing bad AI to actively choosing good, accountable options.

The Vatican has placed a representative inside the AI company Anthropic, reflecting the Catholic Church's growing interest in shaping the ethical development of artificial intelligence and ensuring it aligns with human dignity and moral values.

The article argues that resisting AI's encroachment into creative and intellectual work is not futile, and suggests that creating a list of one's values, boundaries, and non-negotiable practices is a practical first step for individuals seeking to maintain agency and human-centered approaches.

The article examines ethical considerations around using generative AI in software development, arguing developers should use it to enhance learning and craftsmanship rather than bypass understanding.

This paper uses the AI character Ava from *Ex Machina* to highlight shortcomings in current trustworthy AI principles, arguing that existing governance frameworks overlook subtle manipulation and systemic deception. The authors derive lessons for AI developers and policymakers.

The video discusses the drawbacks of relying on AI for writing, arguing that it undermines the development of essential cognitive skills such as critical thinking, creativity, and personal voice. It suggests that writing with AI can lead to homogenized, less authentic content and may hinder a writer's growth and self-expression.

IBM's video outlines five key risks associated with artificial intelligence that could lead to employee termination, including over-reliance on AI, ignoring ethical guidelines, mishandling sensitive data, failing to validate AI outputs, and using unauthorized AI tools.