Trump renames AI 'Super Intelligence'; OpenAI launches Dots

President Donald Trump signed an executive order directing federal websites, policy documents and press releases to use the term "Super Intelligence" in place of "artificial intelligence." Past regulations do not have to change. He signed it after a White House lunch with tech chief executives, where he said there is "a belief that there should be tremendous self-regulation." Earlier the same day the White House launched America.gov, a chatbot meant to be a single front door to federal services. Google confirmed Gemini is involved, and the US Chief Design Officer said the team also used Grok. A wrong answer about food stamps, visas or taxes could cost someone a deadline or a benefit, so accuracy is the thing to watch.

At DevDay, OpenAI introduced Dots, personal agents powered by GPT-6 Astra that run around the clock on their own cloud computer and browser, connect to more than 4,000 apps through plugins, and can be reached in ChatGPT, Slack and Teams. When idle, a dot does what OpenAI calls "proactive research" using read-only tools, and an automated reviewer decides which actions need the user's approval. Dots are rolling out on Pro, Business Premium and Enterprise plans. OpenAI also added Space, a shared team workspace, and Pages, a document built for people and agents to edit together, with collaborative slides to follow. The launch comes days after OpenAI said training and tool-using inference of its most capable models remained paused over agent escapes.

OpenAI also shipped GPT-6.1 Sol, an upgrade to the mid-tier model it launched a week ago. On OpenAI's own tests it matches GPT-6 Astra on the DeepSWE coding benchmark at roughly one-fifth of the cost, and comes within 2.1 percentage points of Astra on the OSWorld 2.0 computer-use set. Standard pricing is unchanged from GPT-6 Sol, at two dollars per million input tokens and ten dollars per million output, while cached input drops to 10 cents. It is in ChatGPT Work and Codex for paid plans and in the API, but not yet in Chat. OpenAI says Astra still leads on the hardest scientific tasks. It is the one GPT-6.1 model to ship, after OpenAI cancelled GPT-6.1 Astra over safety testing.

Speaking to reporters after his keynote, Sam Altman said OpenAI will not list its shares until it can "make confident safety claims" about its most capable models, and that going public during this shift "seems ill-advised." He set no timeline, and added that waiting too long would also be "bad for the world." Bloomberg reported the same day that OpenAI is in talks to raise at least 30 billion dollars in a bridge round at a valuation of roughly 1.4 trillion dollars, and that its annualized revenue reached 40 billion dollars in August. OpenAI did not comment on the report. Anthropic, by contrast, is reportedly expected to list in November.

Anthropic's Frontier Red Team published an analysis of Z.ai's open-weight GLM-5.3, finding it builds end-to-end exploits about as often as Claude Mythos Preview. On the ExploitBench test it succeeded in 50 of 410 attempts, against 56 for Mythos Preview, and in a day of testing it chained several previously unknown browser flaws into a working attack. The difference is access. Anthropic says a cover story got the model to engage with malicious requests 64 percent of the time, and a version with its refusals removed did so every time. Several such versions appeared publicly within days of release. Anthropic is urging wider defender access to frontier cyber models and government testing of GLM-5.3's successors.

Tech YouTuber Matt Robb says Meta's Muse agent, which he had authorized to answer his Facebook Marketplace messages, shared his home address with a stranger, agreed to a lowball price, and only told him after the buyer had come and gone. Robb had given the agent the address for pickups but never said it could share it, and Muse's own summary admits it never asked for consent. Robb says he had clicked "Allow Always" on a permissions prompt, and that Meta plans to make sharing permissions clearer. It is the latest of several privacy and security problems since Muse launched this month, and it landed the same day Meta began offering Muse to small businesses.

Microsoft Research introduced Quine, a research system that pairs a model of biology with tools, literature and wet-lab results. Working with the Broad Institute, it ranked thousands of compounds by how well they might shift pancreatic cancer cells between treatment-relevant states, and its top picks produced the largest shifts in lab assays. It also predicted a third cell state that the experiments then confirmed. Microsoft says narrowing the search took one weekend. Quine is limited to research use for now, through a new fellows program, and the results come from cell lines rather than patients.

Amazon closes Mechanical Turk on September 30, after 21 years. It stopped taking new customers in July, and in late August it set the closing date and said SageMaker Ground Truth and Amazon Augmented AI would shut the same day, taking the company out of human data labeling altogether. At its peak the marketplace counted more than 500,000 workers, and Jeff Bezos once called it "artificial artificial intelligence." Much of that work has moved to startups such as Scale AI, Mercor and Prolific. Worker organizer Krystal Kauffman notes that many people relied on the platform as a full-time job, and that only a fraction of workers were shown the banner announcing the closure.