OpenAI explores reviving Sora and expanding into robotics
A reversal of the retreat from expensive side projects would signal where the company sees its next growth fronts.
Read story
Stories
Tech stories and the sources behind them.
718 stories
Newest first
Sunday
Microsoft to pay Anthropic over $2 billion a year for Copilot
The bill shows how much revenue now flows from major software products to model labs, even as big buyers hedge across providers.
Read storyAnthropic fund and Musk SpaceX pledge duel over AI gains
Two rival plans for spreading AI-era wealth — a company fund that grows as Claude automates jobs, and direct stock giveaways — make who captures the gains a live political question
Read storyiPhone video artifacts raise trust questions about AI processing
A developer documented invented visual details in ordinary footage, a warning for anyone relying on phone video as evidence.
Read storyInvestors push back on Anthropic valuation as revenue growth slows
Prospective investors questioned the numbers in a private meeting, an early test of public-market appetite for AI stocks
Read storyClaude cracks a 25-year cryptography puzzle in half an hour
Solving a bounty problem that stumped cryptographers since its creator died shows how far model reasoning now reaches.
Read storyUnannounced OpenAI models surface in the public API
Stealth listings like this typically foreshadow upcoming releases, and developers are already probing the new entries.
Read storyLinus Torvalds says Linux will never be a political project
The kernel creator's blunt mailing-list stance drew enormous engagement and reignited debate over how open source projects handle social pressure.
Read story
Saturday
Microsoft on track to pay Anthropic over $2 billion a year
The bill shows how expensive renting outside models has become, even as Microsoft shifts more of Copilot to cheaper alternatives to protect margins.
Read storyHackers breach OpenCode and steal developers' AI provider keys
The stolen keys were used to copy models at scale, and defenders found the safety guardrails of their own AI tools blocked them from analyzing the attack.
Read storyCursor probes whether customer data reached SpaceX's AI unit
An internal lawyer warned that training on confidential customer data could create a huge trust problem for a coding tool built on developer secrets.
Read storyReplit names OpenAI its preferred partner, stays model agnostic
The clarification keeps multi-model choice intact on a widely used development platform, easing fears of an exclusive shift.
Read story
Showing 433 to 444 of 718 stories