Epoch AI's FrontierMath benchmark now includes roughly 50 unsolved research-level mathematics problems in its Open Problems track, and AI systems have solved exactly three of them, a 6% success rate. …
Epoch AI, a research organization, reported that parallelization constraints could delay a technological singularity, based on a post shared on July 29, 2026. The finding suggests that limits on paral…
Compute costs could rise 10x or more in coming years, according to an analysis of AI lab economics. If a human-level software engineer could run on an H100 equivalent, that GPU should rent for over $2…
Anthropic's Claude Fable 5 and OpenAI's GPT-5.5 Pro each generated verified solutions to an unsolved number theory problem involving the absolute Galois group of the 2-adic numbers, marking the second…
AI companies are bulk-buying physical books, having them industrially scanned, and then shredding the originals, according to a Futurism investigation. The practice, brokered by data firm ISBNdb under…
Closed-source software no longer protects design secrets because AI can now reconstruct programs by observing their behavior, according to a June test by Epoch AI and METR called MirrorCode. The best …
Open-weight AI models now handle roughly 70% of production tokens routed through OpenRouter, up from 30% a year ago, as enterprises shift from frontier commercial models to cheaper open alternatives. …
Elon Musk predicts artificial intelligence will surpass the combined intelligence of all humans within five years, urging immediate industry-wide safety coordination among frontier AI developers. In a…
The four largest US technology companies are on course to spend close to $700bn on AI infrastructure this year, with combined capital spending set to overtake the cash their core businesses generate, …
Amazon, Alphabet, Meta, and Microsoft are on course to spend about $725 billion on capital expenditure in 2026, up from roughly $410 billion in 2025, while their combined free cash flow is projected t…
Epoch AI tested three leading AI text detectors (Pangram, GPTZero, and Originality.ai) and found that up to 18 percent of AI-generated passages went undetected when language models mimicked an author'…
Open-weight AI models trail closed-weight frontier models by approximately four months in general capability, according to the Epoch Capabilities Index, which aggregates scores from multiple AI benchm…
JS Denain and Anson Ho argue that AI futurism debates overlook the difficulty of developing specific technologies like Dyson spheres or nanotech, focusing instead on AI capabilities. The authors call …
A developer argues that using AI agents to generate code faster is degrading software quality and team alignment, because code was historically a medium for thought and exploration, not just output. T…
The AI Futures Project authors argue in AI 2040: Plan A that the world should agree to an international AI-race slowdown treaty—an arms control deal—that advances alignment and control research while …
AI models retain between high single digits and 40% of customers after five months, with the stickiest foundational cohorts near the top of that range, according to a study by OpenRouter and a16z. The…
Outgoing White House AI advisor Sriram Krishnan stated that President Donald Trump will never support a formal licensing regime for AI, preferring an ad hoc approach with opaque processes. Krishnan ar…
Epoch AI has released an independent database of the world's largest AI data centers, built from satellite imagery, permits, and public documents. The 72 covered sites total 11.2 GW of IT power, led b…
Grid interconnection delays, not electricity shortages, are slowing AI data center construction, with projects like Stargate in Abilene, Texas, facing multi-year waits for utility studies and transmis…
A new model estimates that a 10x reduction in R&D compute for an AGI company would slow AI progress by about 6x in the median case, with an 80% confidence interval of 3.5x to 8x. The model accounts fo…