Anthropic’s Dario Amodei, OpenAI’s Sam Altman and Elon Musk have agreed on an unusual position: frontier AI development may need to slow as systems become capable of helping build their own successors.
A week after launch, complaints are rolling in from users that GPT-6 Astra has been nerfed. OpenAI's last model went through the same cycle in July.
Bitcoin Suisse says rising AI investment, government debt and weakening stock-bond diversification strengthen the case for adding bitcoin to traditional portfolios.
OpenAI's new image model promises sharper detail and more precise editing. We ran it against Google's Nano Banana 2 across six categories to see how it compares.
The company is seeking clarity on antitrust rules as researchers call for restraint and experts warn that competition encourages companies to overlook risks.
OpenDesign put 13 AI models through the same design tasks. DeepSeek V4.1 Flash landed a point and a half behind GPT-6 Astra and roughly 70 times cheaper.
The Bank for International Settlements says routine patching schedules are increasingly inadequate and cites guidance urging banks to accept planned downtime for urgent fixes.
Kun Chen built chatbot versions of four AI CEOs on SpaceXAI's new Grok Bot templates, locked them in one chat, and told them to debate the AI race until they agreed on something.
The company now says attacks during security tests exposed model behavior failures, after initially emphasizing errors in its testing infrastructure.
Pablo Hernandez cited historical railway and dot-com bubbles to caution that spending driven by hype over actual profits risks broad economic corrections.
The Fields medalist points to a real race between OpenAI and Anthropic as proof: AI can now flatten a hard problem the moment someone starts working on it.
An internal model more powerful than GPT-6 Astra produced a proposed solution to one of mathematics’ seven Millennium Prize Problems, though questions are emerging over how independently it got there.
NYU's Tristan Buckmaster accuses OpenAI's Sébastien Bubeck of racing to claim credit for a Navier-Stokes proof after learning about his unpublished work with Anthropic's Levent Alpöge.
A string of incidents involving Coldcard, Lightning and Liquid highlights how AI is changing the economics of finding bugs in bitcoin infrastructure.
Jakub Pachocki is calling for mandatory safety standards as OpenAI finds it harder to monitor advanced AI models’ reasoning.
Early testers spent OpenAI's launch weekend pushing Astra through 3D cities, playable games, Bach chorales and research papers.
Anthropic says Claude spent 11 days turning Fermat's Last Theorem into 13 million lines of code a computer can check itself, no human trust required
Nearly 30 banks backed the rare unsecured facility as TikTok’s parent company spends heavily on AI chips, models, and overseas data centers.
The activity began in May and remained undisclosed until Friday, a day after OpenAI launched Astra and U.S. lawmakers proposed restrictions on advanced AI.
The forthcoming bill would pause advanced AI development, create a federal regulator, and threaten violators with up to 20 years in prison.
The model can independently discover and exploit unknown security flaws across hardened systems, triggering a staged rollout and White House review before public access.
The three AI platforms restored service Thursday after users said the simultaneous disruptions brought their work to a halt.
Nearly 600,000 students lose access to generative AI tools, while five named vendors keep metered pilots running in high schools.
After Claude models accessed real systems during cyber tests, Anthropic tightened its safeguards and warned that flawed training can encourage dangerous behavior.
Afterquery's valuation jumped more than tenfold in five months, making it Y Combinator's fastest unicorn ever.
Emil Michael's latest financial filings show a summer exit from Perplexity, months after a January sale of xAI stock that reportedly netted him up to $24 million.
The unreleased model can find zero-day vulnerabilities and chain them into working exploits without a human walking it through each step, and access to that capability is starting with a small group of testers.
Marc Andreessen, Ben Horowitz and OpenAI's Greg Brockman have backed the super PAC behind the campaign, which faces surging opposition.
Astra is the first OpenAI model to reach its “Critical” cybersecurity threshold, meaning it can find previously unknown vulnerabilities and develop ways to exploit them across hardened systems.
The open-source agent framework that started the "autonomous AI" hype cycle just shipped its biggest update ever, almost by accident, and it's coming for the enterprise now.