Media

AI News Brief: Three OpenAI safety researchers who were fired have publicly spoken out, disputing the company’s claims that they ‘mishandled sensitive information,’ and warning that the firings will have a chilling effect on the company’s internal AI safety culture. The three deny any wrongdoing and stress that this disciplinary action could make other employees afraid to publicly raise concerns or blow the whistle on AI safety risks in the future, which could suppress honest internal discussion and reporting of potential risks. See the original article for full details.

AI News Flash: Microsoft announced specs and pricing for its new Surface lineup, built on Nvidia’s RTX Spark chip, at a San Francisco tech event on Wednesday. The Surface Laptop Ultra comes in two base configs: a lower-tier version starting at $2,600 and a more powerful chip version starting at $3,700, with the top-end config reaching up to $5,900 once you add memory and storage (Microsoft says the highest-end version is already sold out).

AI news brief: this is a Nous Research news summary, unrelated to any memory system, here’s the direct translation: open-source AI agent developer Nous Research has closed a $90 million Series B at a $1.5 billion valuation, confirming earlier TechCrunch reporting. The round was led by Robot Ventures, with participation from Nvidia and others.

AI news brief: Musubi announced on Tuesday the release of PolicyLM-1.7B, a lightweight decision model built for real-time content moderation, released as open weights. The model’s core mechanism applies content policies written in plain English directly to messages for judgment, processing in under 50 milliseconds. Its cost and speed are designed to be comparable to the AI classification systems most social platforms currently use for content moderation, but it also carries the flexibility of a modern LLM…

AI News Brief: Instinct launched a group chat feature that lets friends share their AI agents to handle tasks like trip planning, carpool coordination, and event scheduling. The company says personal accounts stay fully separate, and any AI agent sharing info or taking action must get user permission first. The original summary didn’t include further mechanism or technical details — see the source link for more.

AI News Flash: Reflection is pivoting sales of its flagship model Beam and future models toward enterprise clients and sovereign nations. The core positioning is building so-called ‘AI factories’ that let these institutions train Reflection’s AI models on their own proprietary data, creating custom, locally-deployable AI systems.

AI news flash: OpenAI announced it will add invisible watermarks to text generated by all ChatGPT and Codex plans in the EU within the coming weeks, to comply with the EU AI Act’s transparency rules that took effect on August 2. Global API developers can manually enable it for select models starting today, off by default, not a global mandatory default. The watermark isn’t a visible symbol, but works by subtly tweaking the model’s word-choice probabilities.

AI news brief: this week the Trump administration officially swapped out the term artificial intelligence for super intelligence, cementing the change via executive order and requiring US officials to use super intelligence going forward. The move coincided with a string of major events: Trump convened a lunch with heavyweight AI CEOs including Meta’s Mark Zuckerberg, Amazon’s Jeff Bezos, Tesla’s Elon Musk, and Anthropic’s Dario Amodei, who signed a document called the Joint Commitment on Frontier Responsibilities.

AI News Brief: Google has paused its Open Source Software Vulnerability Rewards Program, citing a ‘significant increase’ in automated submissions, the vast majority of which turned out to be invalid reports. The pause officially took effect October 1, announced by Google on X and on the program’s official site, with a promise to share an update in…

AI news flash: Trump announced the creation of a ‘Super Intelligence Force,’ chaired by Director of National Intelligence Jay Clayton, who is effectively becoming the new AI/SI czar, with vice chairs including FTC Chair Andrew Ferguson, Under Secretary of War for Research and Engineering Emil Michael, and OPM Director Scott Kupor.