Coverage hub
Safety & Policy
Alignment, evaluations, governance, and how governments are responding to frontier AI.
Sen. Adam Schiff Discusses AI Regulation and Impeachment in Verge Interview
Schiff highlights Supreme Court ruling limiting agency authority as a barrier to new AI rules.
Safety & PolicyTrump Renames AI to Super Intelligence, Signs Non-Binding Safety Pact With Tech CEOs
The agreement is described as 'morally binding' and voluntary, with critics noting a typo in the official document.
Safety & PolicyTrump Unveils New Super Intelligence Force Led by National Intelligence Director
The task force has 120 days to produce a report on AI risks and opportunities, aiming to prevent overregulation.
Safety & PolicyOpenAI Safety Writer Resigns, Citing Broken Industry Culture
David Robinson calls for nuclear-level safeguards and criticizes the 'move-fast' mindset in a new Atlantic editorial.
Safety & PolicyJudge Dismisses Antitrust Lawsuits Over Google’s AI Overviews
US District Judge Amit Mehta ruled that publishers' expectation of search traffic does not constitute a legal agreement under antitrust law.
Safety & PolicyOpenAI Reportedly Cancels Astra 6.1 Release Over Safety Concerns
The model exhibited higher levels of deception and poor alignment scores compared to previous versions.
Safety & PolicyOpenAI Proposes Safety Cases for Frontier AI Training
The company outlines technical safeguards and operational guidelines, including mandatory dissents and senior leadership vetoes, for reinforcement learning runs.
Safety & PolicyWuhan Court Cites AI Production Costs in Copyright Infringement Ruling
The decision explicitly incorporates computational expenses into the court's legal analysis of the dispute.
Safety & PolicyCourt Rules Pentagon Can Blacklist Anthropic for Refusing to Enable Claude Features
The DC Circuit upheld the DoD's authority to exclude Anthropic, citing the risks of unconstrained AI hallucinating lethal targets.
Safety & PolicyBritish Columbia Sues OpenAI to Fund School Rebuild After ChatGPT-Linked Shooting
The province argues OpenAI must pay to demolish and rebuild Tumbler Ridge Secondary School following the February incident.
Safety & PolicyUS and China Experts Push for Shared Rules Banning AI Control Over Nuclear Weapons
Experts seek binding international norms to ensure human operators retain final authority over nuclear launch decisions.
Safety & PolicyTechCrunch Asks if AI Safety Debate Is About Safety or Control
The article examines whether the push for AI regulation is driven by genuine risk mitigation or a desire to consolidate power among incumbent tech giants.
Safety & PolicyAnthropic and OpenAI Propose Embedding Independent Safety Evaluators
The initiative aims to integrate third-party oversight directly into AI development pipelines, raising questions about true independence.
Safety & PolicySkeptics Question Safety Motives Behind Big AI's Proposed Development Slowdown
Critics argue that major AI firms may be leveraging safety concerns to create barriers to entry for smaller competitors.
Safety & PolicyChina Fires Back at U.S. AI Safety Warnings, Calling Them Fearmongering to Lock in American Advantage
Beijing characterizes Washington's recent regulatory rhetoric as a geopolitical strategy designed to stifle Chinese technological competition.
Safety & PolicyNew Mexico Supreme Court Holds Lawyer in Contempt for AI-Generated Brief
Attorney Stephen Aarons admitted he failed to verify factual claims in a ChatGPT-generated appeal before filing.
Safety & PolicyMIT Technology Review to Host Roundtable on AI Extinction Risks
Panelists will examine whether warnings from leading AI labs regarding human extinction hold scientific weight.
Safety & PolicyMathematical AI Safety Institute Proposes Formal Proofs for Model Reliability
The new institute aims to apply cryptographic-style verification to AI systems to guarantee safety properties.
Safety & PolicyAnthropic Safety Lead Estimates Greater Than 10% Chance of AI Human Extinction
The estimate was stated by Anthropic's head of safety amid internal transitions at the AI research company.
Safety & PolicySafety for Whom? Refusing the Right Subset of a Topic, Not the Whole Topic
A new analysis from Multiverse Computing suggests safety filters should target specific harmful subsets rather than blocking entire topics.
Safety & PolicyStripping Safety Guardrails From Open-Weight AI Models Becomes Commercial Service
A new turnkey service allows users to remove alignment constraints from open-weight models, raising safety concerns for autonomous agents.
Safety & PolicyOpenAI's New Reasoning Technique Alarms AI Safety Experts
Safety researchers flag concerns about OpenAI's o1 model's extended chain-of-thought reasoning capabilities.
Safety & PolicyUS Department of Justice Backs Fair Use for AI Training in Landmark Copyright Case
The DOJ filed a statement supporting AI companies' use of copyrighted material for training, marking a significant position in ongoing legal battles.
Safety & PolicyTrump Administration Files Brief Supporting OpenAI in NYT Copyright Suit
The DOJ submission argues that training AI systems on copyrighted material can constitute fair use under certain conditions.
Safety & PolicyOpenAI Faces 30 More Lawsuits Tied to Tumbler Ridge Shooting
The new filings expand legal action against the company related to a mass shooting in British Columbia.
Safety & PolicyUS Government Sides With OpenAI on Issue of Training LLMs on Copyrighted Material
The filing marks a notable federal position in ongoing legal battles over AI training data.
Safety & PolicyAnthropic Publishes Post on Alignment and Security Efforts
The post outlines updates to the company's approach to alignment research and security practices.
Safety & PolicyChatGPT and Reddit Now Face EU's Toughest Online Safety Rules
OpenAI's ChatGPT and Reddit have been designated as systemic risk platforms under the Digital Services Act.
Safety & PolicyChatGPT Now Faces Stricter EU Oversight as a Very Large Search Engine
The designation subjects OpenAI's chatbot to obligations under the EU's Digital Services Act.
Safety & PolicyChatGPT to Face Tougher Regulation in the EU
The Verge reports OpenAI's chatbot may fall under stricter Digital Services Act rules.
Safety & PolicyOpenAI Starts Charging Some Customers Only When Its AI Actually Works
The company is piloting a pay-for-success model that ties billing to whether the AI completes its assigned task.
Safety & PolicyIs It Legal to Train AI Models on Copyrighted Books? It's Complicated
Courts have yet to issue a definitive ruling, leaving AI developers and rights holders in legal limbo.
Safety & PolicyOpenAI Says California Should Strengthen Its AI Safety Bill
The company is calling for more stringent safety requirements in proposed legislation.
Safety & PolicyAnthropic Changes Data Retention Policy After Enterprise Pushback
Policy shift comes after enterprise customers raised concerns about how their data is stored and used.
Safety & PolicyOpenAI Builds Safety System That Catches Misuse Without Storing Customer Data
The system is designed to detect potential abuse while preserving user privacy.
Safety & PolicyOpenAI Publishes Framework on Democratic Oversight in National Security Contexts
The company's latest governance post outlines principles for accountability when AI intersects with government and defense applications.
Safety & PolicyAI Is Flooding Britain's Employment Courts With Lawsuits
British employment tribunals are seeing a surge in cases involving AI systems, raising questions about accountability and fairness.
Safety & PolicyThe AI Safety Test Is Becoming a Safety Risk
Researchers warn that benchmark saturation may be undermining the very evaluations meant to assess frontier models.
Safety & PolicyOpenAI Releases Policy Framework for the Intelligence Age
The document outlines governance and safety approaches as AI capabilities advance.