Anthropic is poised to prove an AI lab can be profitable, but if progress slows, low switching costs and distillation could force labs to compete on price rather than capability, turning intelligence …
METR contributor Ivan Bercovich argues that most AI benchmarks are flawed and that building good ones requires nuanced understanding, drawing on 18 months of experience with Terminal Bench. Good tasks…
Athena Insights launched Americans on AI, a tracking survey of how American adults feel about artificial intelligence, fielded every two weeks by NORC at the University of Chicago on a probability-bas…
A critical review of AI 2027 claims that the report misrepresents scientific evidence, specifically regarding AI reasoning and AI training AI. The reviewer argues that a referenced paper on reinforcem…
A new paper from researchers argues that AI safety discussions must prioritize preventing large-scale suffering caused by misused or misaligned AI, as well as from uncontrolled AI aligned with values …
The Sentient Futures Project Incubator is accepting applications for its Fall 2026 cohort, a 10-week program running from August 31 to November 9, with a deadline of August 9, 2026. The program offers…
Australia will not proceed with mandatory AI guardrails, instead relying on existing voluntary standards and a new AI Safety Institute, according to the December 2025 National AI Plan. Security practi…
Pausing artificial intelligence development at human level is harder than pausing as soon as possible, argues Michael Dickens in a cross-posted essay. Dickens cites economic incentives to continue, AI…
A new analysis warns that conservation-focused population estimates for wild animals, which often count only mature individuals, can severely underestimate the number of animals affected by welfare in…
A content creator conducted street interviews across the US, finding that most Americans are unaware of AI advancements and have not used frontier models, yet they are directionally concerned about AI…
A student planning to study dentistry while pursuing AI safety research is weighing whether to spend three months preparing for a more competitive medical university that offers better access to a top…
PauseAI opened applications for PauseCon London '26, a training conference for organizers working to prevent unsafe AI development, scheduled for September 5-7 in London. The event offers free partici…
Public health physician Juliana Grant is seeking feedback and collaborators for a proposal to create a multi-lineage AI model panel that reviews safety evaluations for welfare concerns. The panel woul…
The AI Futures Project released a detailed vision for international AI regulation centered on total research transparency, arguing that open access to all AI research would simplify governance and enf…
AI governance faces a critical test as geopolitical rivalries threaten to undermine emerging international frameworks, according to Cyrus Hodes of AI Safety Connect. Despite a proliferation of summits…
The team behind AI 2027 released a new scenario called AI 2040: Plan A, offering a positive vision for navigating the creation of super-intelligence. The named authors include Thomas Larsen, Romeo Dea…
Sneha Revanur, founder of AI advocacy nonprofit Encode, helped pass landmark AI safety laws in California and New York by building an unlikely coalition of activists, despite facing opposition from Bi…
Economists are increasingly studying the risks of transformative AI, including potential human extinction, and this post provides a guide for economics researchers to identify high-impact research are…
AI safety grantmaking suffers from combining three distinct jobs—field diagnosis, project design, and applicant selection—into a single proposal process. They highlight two alternative models: the Ins…
A researcher applied METR's time-horizon methodology to Microsoft Excel and found it completes tasks requiring 6.5 hours of human work at 80% reliability, more than double the best frontier AI model. …