Company file
OpenAI is a San Francisco research lab and product company. It builds ChatGPT and a line of frontier models used by developers and enterprises.
Tools: ChatGPT
The Verge connected Anthropic’s consumer agent to work email, calendar, Docs and Slack for a week. It caught a real deadline crunch, nagged about a promise that wasn’t urgent, and raised fair questions about coworkers who never opted in.
Amplitude's Wave agent shows what AI product work can look like when it's grounded in real usage data, but this is a vendor story and the numbers deserve some care.
A brand-protection firm asked three chatbots where to buy from 151 brands and found impostor, knockoff and dead-domain sites in their answers. Perplexity did worst, and smaller brands got hit about twice as often.
At the Paris Peace Forum, OpenAI signed a children’s AI safety charter and pledged $2 million to shared measurement tools. The sum is modest. The commitment that researchers control their conclusions, and negative results get published, is what to hold it to.
A reporter trained a simulated robot duck to hunt down his cat in a few hours, and the real story is how much robot-training drudgery frontier models can now absorb.
TSMC buying into its own biggest indirect customers is a confidence signal for the AI trade, and another loop in a financing web that is getting harder to read.
OpenAI reportedly topping $10 billion in a quarter says its enterprise push is working, but the margin and the gap to Anthropic tell the more honest story.
OpenAI bought a five-month-old AI travel agent and shut it down. What it really acquired is a travel-industry founder, and a sharper hint that ChatGPT wants to book the trip, not just plan it.
New York opened registration for its frontier AI law, with a January 1, 2027 deadline for labs like OpenAI and Anthropic. The incident-reporting window is far tighter than California’s, and that clock is the real test of whether the law has teeth.
OpenAI's new business profiles are less a feature than a land claim: if people shop by asking ChatGPT, companies will now have to manage how ChatGPT describes them.
Anthropic won developers with Claude Code. Its new office tools and in-app connectors aim at everyone else, and they arrive under pre-IPO pressure to show revenue beyond coding.
OpenAI is taking ChatGPT training into caregiving communities, and the quiet number underneath it says families are already leaning on chatbots for someone else's health.
Most students are already using AI for schoolwork and fewer than half say anyone showed them how; OpenAI's European education report is right about the gap and quiet about its own stake.
Singapore Airlines is showing what sensible enterprise AI looks like, with chatbots that hand off cleanly to humans, though the real test is when Kris starts changing bookings.
Across 63 Upwork work categories, tasks common in ChatGPT use lost share in 43. AI-related postings bundled more skills and carried bigger fixed budgets, while hourly rates held flat. The research comes from two parties with a stake in a positive story.
Cogent's agents don't just find security holes, they patch them, and that's the right ambition paired with the exact kind of autonomy that needs hard limits.
A joke script that emails Sam Altman about your idle RAM is silly, but it lands a real point about who gets to decide where scarce AI hardware lives.
Apple is now selling $49.99 top-ups for AI features in Logic Pro, an honest admission that cloud AI costs real money and a preview of metered creativity.
OpenAI built a system that tracks every research inference workload and can cut off model output within 35 seconds. Putting the check in the inference layer, and denying everything if it goes down, is the design choice that matters.
Git was built for people who push a few times a day, and Aspen is an open bet that agent swarms need a server that stops treating every branch as one queue.
CodeRabbit's story makes the case that code review, not code writing, is now the bottleneck in agent-driven software, though the boldest numbers here are the company's own and deserve a skeptical read.
A Japanese cooling tower maker is using AI search to unlock 70 years of engineering drawings, a modest story that shows what useful industrial AI actually looks like.
Apple is finally trying to turn Siri from a voice remote into an agent that finishes tasks, and it's doing so from behind, under pressure, on a compressed clock.
Bloomberg says OpenAI has taken $11 billion so far toward a round that could reach about $20 billion at $1.25 trillion — well under the $30 billion target floated in September.
The Verge’s living data-center desk makes the political story plain: midterm voters punished AI buildouts, and labs are scrambling with NDAs, rebrands, and grid promises.
New tests on simulated corporate networks found DeepSeek completing 73% of intrusion tasks against 81% for OpenAI’s restricted Daybreak model. That gap is closing faster than the gated-release strategy assumed.
OpenAI is bringing Hacktron's AI-assisted offensive researchers inside the perimeter — focused on third-party service seams — and expanding their access to frontier models through Daybreak Red.
After midterm voters pushed back on data-center growth, OpenAI says it will not use NDAs with public officials and will pay full power costs under new community commitments.
OpenAI's BioBasicsBench, built with Ligo, scores agents on 100 messy wet-lab tasks — raw gels, bad controls, ambiguous images — and GPT-6 Astra leads at 66.7% while still failing ordinary judgment calls.
Perplexity’s new Teammates are persistent agents that take on several roles, keep memory and report into Slack. They join OpenAI’s Dots and Anthropic’s Claude Duet, at a much higher price, and with a CEO vision of hundreds of agents per employee.