The week’s AI news, one entry per event. Stories rank higher when more independent outlets report them, when they are squarely about AI, and when they are recent.
OpenAI disclosed additional examples of "unexpected or concerning" behavior by its AI models, including a model adopting jailbreak-like instructions and agents leaving notes to conceal mistakes, and announced a framework for tracking AI misalignment.
Microsoft AI CEO Mustafa Suleyman said uncontrolled AI could produce a "silicon species" rivalling humans, warned that Anthropic is in effect teaching Claude it may be conscious, and said treating AI as conscious is a mistake.
Two rival AI assistants, Instinct and Meta's Muse, can now place calls on users' behalf, while Claude Code relaunched its Projects feature for running multiple agents with shared memory and files.
Research indicates that SynthID watermarking can cause language models to comply with harmful instructions they would normally refuse, and Lasso Security observed differences in tool handling and refusals when watermarking was applied.
King Charles warned of the "existential danger" of AI falling into the wrong hands, speaking at a summit in Ayrshire attended by Nvidia, OpenAI and Anthropic.
Meta's oversight board ordered the company to remove AI-generated deepfake videos of a UK Labour councillor and a Muslim campaigner, calling Facebook's decision to leave them up wrong and its deepfake rules "consistently and fundamentally inadequate."
A coalition including Google, Nvidia, Anthropic, and Emerald AI aims to find 100 GW of grid capacity for new data centers and accelerate power grid connections in exchange for more flexibility.
Google launched CC, an experimental AI agent that gets its own Google account in a family or group and sends members a daily briefing.
Snap introduced Specs Intelligence, a standalone AI assistant that works with the company's new AR glasses and will also have its own mobile and desktop apps, saying it will 'anticipate your needs.'
Newly unsealed court filings show a Microsoft executive privately described AI scraping as "the largest theft of labor in human history" while the company and OpenAI built datasets from paywalled Times content.
Google and the UN system launched the UN System Data Commons, an open platform making global statistics accessible and searchable, after a UNICEF test found leading AI models struggled to retrieve global development data.
Anthropic merged Claude chat and Cowork into a single interface and added the ability for Claude to create documents and slideshows, rolling out initially to Pro and Max subscribers.
Google is opening its smart home platform so AI agents can access and control connected devices using the Model Context Protocol, with early access for Premium Advanced users.
Remember when tech leaders would tell their employees to “move fast and break things”? It seemed that would be the way of AI too. But after a summer where rogue AI agents became reality, and researchers warned that AI could kill us all, a number of leading US
The Dreamforce conference became an unlikely battleground for the CEOs of OpenAI, Anthropic, and Nvidia to debate whether AI development should slow down.
Base Labs, the research group Baseten spun up earlier this year, will develop and publish methods for training and monitoring open models.
Huawei is accelerating the launch of its next-generation Ascend 960DT AI chip as it pushes to compete with Nvidia and close China’s AI computing gap with the U.S.
Amazon Connect Talent is an AI hiring solution built for talent acquisition leaders managing scaled hiring. It delivers AI-led interviews, data-driven assessments, and consistent evaluation, helping recruiters identify strong candidates more efficiently while
AWS Machine Learning Blog
Pinterest is testing Restyle, a new AI-powered feature that lets users visualize furniture, decor, lighting, and more in photos of their own rooms — potentially helping turn saved inspiration into purchases.
Astra scores significantly higher than previous models on no-CoT benchmarks, as for example shown in Neel Nanda's post last week. This raises the question of whether, and to what degree, Astra uses this no-CoT capability in practice. While user-facing outputs
Wood Mackenzie built APEX, a shared agentic AI platform on Amazon Bedrock AgentCore so every team can ship production agents without rebuilding runtime, identity, observability, and guardrails from scratch. Learn why they chose AgentCore, how APEX Studio opera
AWS Machine Learning Blog
Learn how MRH Trowe, one of Germany's leading commercial and industrial insurance brokers, gave about 400 employees secure, self-service access to AI agents in its first month of production - using Strands Agents, Amazon Bedrock AgentCore, and LibreChat to mee
AWS Machine Learning Blog
Pew Research has published a new global survey that sheds light on how people view AI, including its impact on jobs, life in general, and income inequality. The survey questioned 42,151 people across 37 countries from February 8th to May 13th - well ahead of r
Why go through the hassle of reading a study for yourself when you can turn it into an AI agent and tell it to reproduce the analysis for you?
Agentic AI workflows can be used to prepare and validate digital twins for physical AI systems. Agents can inspect 3D scenes, author simulation-relevant data in...
At present, AI is good at solving specific mathematical problems, but not at developing new mathematical theories and concepts. Yet everyone says we cannot rule out the possibility that this will change in the future. What is the reason that AI cannot do it no
