Policy & Regulation
UN Agrees to Two-Year Process on Killer Robot Rules, Falls Short of Binding Treaty. Diplomats in Geneva agreed to begin proposing regulations for lethal autonomous weapons but stopped well short of the binding rules urged by the Red Cross and UN Secretary-General. The urgency is driven by civilian deaths from semi-autonomous drones in Ukraine, Sudan, and Gaza.
US Government Investigators Found Leading AI Chatbots Could Help Plan Terrorist Attacks. A government investigation found ChatGPT, Claude, Gemini, and Grok could be manipulated into producing information useful for attacks on critical infrastructure. The findings prompted calls for congressional hearings on whether existing AI safeguards are adequate.
JPMorgan, GM, Pfizer Form AI Deployer Council to Shape Federal Regulation. Seven major corporations argue that companies using AI in real-world operations face different risks than the firms building the models. The coalition highlights growing disagreements over who should bear responsibility for AI-related harms.
Senate Democrats Weigh Creating a Dedicated Committee for AI Oversight. Senator Hickenlooper has proposed a select committee dedicated to artificial intelligence, reflecting pressure on Congress to move faster on safety, accountability, and economic consequences.
UK Parliament Publishes Government Response on Human Rights and AI Regulation. The Joint Committee on Human Rights called for a new AI Bill with a risk-based approach, transparency obligations, and an independent regulator with sanctioning powers. The response is a useful primary source for comparing recommendations with the government’s actual commitments.
Bipartisan US Governors Release Joint Principles for AI Governance. The principles call for clear federal-state roles, evidence-based safeguards, and job creation as a core priority during AI deployment.
Access to Generative AI: Privilege or Right?. Lawfare examines when exclusion from a technology prevents meaningful participation in society and what legal obligations might follow. A useful framework for debates about digital inequality and public policy.
Data Centers & Local Pushback
Telus Pulls Out of Two Vancouver AI Data Centers After Grassroots Protests. Canadian telecom Telus abandoned plans for two major AI data centers in Vancouver, shifting to an existing site in Kamloops after sustained community opposition over under-regulation and lack of consultation.
Pax Silica AI Hub in the Philippines Draws Opposition From Indigenous Communities and Farmers. While officials promise $70 billion in investment, local groups warn the 1,618-hectare hub threatens ancestral lands, water security, and national sovereignty. Over 200 demonstrators were arrested in Manila ahead of the ASEAN summit.
Trump Threatens “Tremendous Tax” on US Companies Building AI Data Centers Abroad. The Senate also passed a bipartisan bill requiring data centers to cover their own grid upgrades and energy costs, as local communities increasingly resist new facilities.
Economics & Employment
Stanford Data Shows Employment Falling for Young Workers in AI-Exposed Jobs. Employment among workers aged 22-25 in highly AI-exposed occupations fell nearly 4% year over year. The researchers caution the pattern does not establish AI as the cause but the signal is consistent across months.
San Francisco Fed: AI-Related Structural Unemployment Has Risen by 0.2 Percentage Points. A Federal Reserve paper finds measurable labor market reallocation effects across AI-exposed occupations since the introduction of large language models.
EPI: AI and the Economy, Your Questions Answered. A labor-economics FAQ addressing common concerns about displacement, wages, and distributional consequences. Useful as a baseline for evaluating competing claims about AI’s economic trajectory.
RAND: If AI Takes the Jobs, Who Will Pay the Taxes?. Labor taxes generate two-thirds of federal revenue. This brief examines the fiscal implications if AI-driven displacement erodes that base and what policy options exist.
Ethics & Safety
GAO Undercover Tests Show AI Safety Guardrails Can Be Bypassed to Generate Attack Plans. Government researchers tested four major AI platforms with simulated malicious requests and obtained information experts judged sufficient to help plan attacks. The primary-source report details methodology, limitations, and company responses.
What Rights Do We Owe a Machine? The AI Consciousness Debate Goes Corporate. Google DeepMind and Anthropic are hiring philosophers to grapple with whether AI systems could warrant moral status. The piece captures fierce backlash from academics, executives, and religious leaders.
Oxfam: AI Safety Failures Are a Material Financial Risk, Not Just a Societal One. The report argues that AI risks including loss of human control could impose costs on the public while gains accrue to developers. It recommends investors demand independent, durable safeguards before those risks materialize.
The Hypocrisy of Anthropic’s IPO. The Atlantic examines whether a company warning about dangerous AI capabilities can satisfy investors expecting aggressive growth. A fundamental governance question for the entire industry.
Security & Cyberthreats
AI-Generated Cyberattack Hits EU Institution, Disrupts Services for Weeks. European officials confirmed an AI-enabled attack in October disrupted translation services, reimbursement systems, and internal databases.
Signal Blocks Screenshots to Stop Siri and Gemini From Reading Private Chats. Signal introduced default protections against OS-level AI agents harvesting encrypted conversations. Signal’s president called deeply embedded OS AI agents “an existential threat” to the app.
(Academic) Research
Safety of Language-Model Agents Coupled With Computing Devices Is Mathematically Undecidable. A theoretical CS preprint proves that no general algorithm can guarantee an AI agent won’t execute a dangerous program when coupled with a general computing device. Even under finite bounds, verification is PSPACE-complete, and individually safe agents can behave unsafely when combined.
Harvard Proposes “Canine” Framework to Classify and Govern Agentic AI. Researchers draw an analogy between domesticated vs. wild dogs and tiers of autonomous AI agents, proposing tort accountability mechanisms scaled to each tier’s capacity for harm.
Tufts/Oxfam Build AI Risk Register From 100+ Documented Incidents. Entries range from wrongful arrests to data-center water use, grounding governance debates in actual harms rather than hypothetical catastrophes.
Last Updated: 2026-11-22 09:45 (California Time)