US House lawmakers release draft bill to prohibit state AI rules ↗
A bipartisan House draft would block states from regulating how AI models are developed - though states could still regulate how AI is used.
Tech groups welcomed the national-standard idea. Consumer advocates, much less so. They warned it could leave bias, fraud, deepfakes and market power rather undercooked.
Anthropic says AI labs need coordinated plan to halt development if risks rise ↗
Anthropic said frontier labs need a verifiable way to slow or pause AI development if systems start improving themselves too quickly.
The company argued that a solo pause would mostly hand momentum to another lab. It is a safety pitch, but also a game-theory knot.
Meta repeatedly pushes back new AI model release for developers, WSJ says ↗
Meta’s Muse Spark API has reportedly slipped again, with no firm launch date at the time of the report.
Meta says early partners are testing it and a release is still expected this month. So it is delayed, but not dead - at least, that is the current read.
Dreaming: Better memory for a more helpful ChatGPT ↗
OpenAI began rolling out a new ChatGPT memory system built on “dreaming,” aimed at making memory fresher, more scalable and less stale.
Plus and Pro users in the US get it first, with Free and Go users coming later. The big claim: memory becomes less like sticky notes and more like a living scrapbook… hopefully one with manners.
Biodefense in the Intelligence Age ↗
OpenAI published an action plan for AI-powered biological resilience, tied to its GPT-Rosalind work in biology and drug discovery.
The argument is blunt: powerful bio-AI can help medicine, but it also raises security risks. OpenAI wants trusted defenders equipped before the scary stuff shows up in the hallway.
Nemotron Coalition ↗
French computer-use AI startup H Company joined NVIDIA’s Nemotron Coalition, an open-source frontier model initiative.
H is positioning its Holo3 and HoloTab products around agentic computer-use - AI that does things inside digital interfaces, not just chats prettily in the corner.
FAQ
What does the US House draft AI bill aim to change?
The draft bill would create a national approach by preventing states from regulating how AI models are developed. States could still regulate how AI is used. Supporters see this as a way to avoid a patchwork of state rules, while critics worry it may weaken protections around bias, fraud, deepfakes and market concentration.
Why are some AI companies asking for coordinated pauses in development?
Anthropic argues that frontier AI labs may need a shared, verifiable plan to slow or halt development if risks rise quickly. The concern is that one company pausing by itself could simply allow another lab to move ahead. A coordinated pause is framed as both a safety measure and a practical response to competitive pressure.
What happened with Meta’s new AI model release for developers?
Meta’s Muse Spark API has reportedly been delayed again, with no firm launch date at the time of the report. Meta says early partners are still testing it and that a release is expected this month. For developers, the key point is that the product appears delayed rather than cancelled.
How is ChatGPT memory changing with OpenAI’s “dreaming” system?
OpenAI says its new ChatGPT memory system is designed to keep memory fresher, more scalable and less stale. The rollout begins with Plus and Pro users in the US, with Free and Go users coming later. The goal is to make memory feel less like static notes and more like an evolving personal context.
Why is OpenAI talking about biodefense and AI now?
OpenAI published an action plan focused on AI-powered biological resilience, linked to its GPT-Rosalind work in biology and drug discovery. The article frames bio-AI as valuable for medicine, while also noting potential risks. OpenAI’s position is that trusted defenders should have strong tools before biological security threats become harder to manage.
What is the Nemotron Coalition and why did H Company join it?
The Nemotron Coalition is NVIDIA’s open-source frontier model initiative. H Company, a French computer-use AI startup, joined the coalition while positioning products like Holo3 and HoloTab around agentic computer-use. That means AI systems designed to act inside digital interfaces, rather than only respond in chat.