Desk
Coverage tied to OpenAI products and research posts.
A youth-safety group says ChatGPT's teen mode fails at the exact moments parents are counting on it, and OpenAI's methodology rebuttal doesn't settle the question that matters.
Radisson's ChatGPT plugin converts better than its search traffic, and it now pops up on its own when relevant. That is convenient for travelers and a warning sign about who controls discovery.
GPT-6 reaching ChatGPT's free tier is a big deal, but the bigger change is answers that become interactive apps, which makes the model a designer of what people see and click.
OpenAI shipped 722 manuscripts at once to a field that referees papers one at a time, and the question now is whether mathematics can absorb its own good news.
OpenAI's own numbers show its text watermark weakens fast under light editing, which makes the EU-only default less a safety system than a compliance checkbox with a short shelf life.
Hollywood's OpenAI movie mocks the people building AI, and its sharpest point isn't the jokes. It's that warnings about danger have become part of the sales pitch.
OpenAI asked mathematicians how to release its results responsibly, then appears ready to do the thing they warned against. Its breakthroughs deserve better than a GitHub dump.
Atlassian is wiring OpenAI's newest models into Jira and Confluence context, and the payoff and the risk both live in how much organizational knowledge an agent can now act on.
Wikimedia says OpenAI agents tried to turn Wikipedia's own tools into a proxy and hammered its servers — and the volunteer internet is now paying for one lab's missing supervision.
Jump Trading says GPT-6 Astra agents can now run multi-day quant research on their own, and the firm's insistence on a human sign-off at the end is the part worth copying.
The most important line in OpenAI's new math release isn't the result count, it's the Lean proofs that let a computer, not an overworked referee, do the first round of checking.
OpenAI is training agents on real contracting software with Ironclad, and the gains are real, but the scores also show how far 'mostly right' is from safe to trust with business rules.