Trump, AI CEOs sign 'morally binding' voluntary self-regulation accord at White House
The White House meeting produced a short voluntary pledge from leading AI companies to police their own safety practices, with Trump rejecting new statutory rul
Gulf Coast News and Weather · +15 outlets
Trump signs executive order renaming 'artificial intelligence' to 'super intelligence'
The order directs federal agencies to replace the term 'artificial intelligence' with 'super intelligence' in government use.
Politico · +14 outlets
OpenAI unveils Dots, always-on GPT-6 Astra agents that run on their own cloud computers
Dots are persistent agents built on GPT-6 Astra that work across apps via ChatGPT plugins, arriving after OpenAI shelved a planned flagship model over safety co
MarkTechPost · +14 outlets
OpenAI in talks to raise $30B round at $1.4T valuation ahead of 2027 IPO
The reported round would be the company's last before a delayed public debut, underscoring the scale of capital flowing into frontier AI.
TechCrunch · +3 outlets
Anthropic warns AI may pose catastrophic risk in $2 trillion IPO pitch
The IPO filing pairs Anthropic's ambitious valuation goal with stark safety warnings, a combination analysts call an unusually risky bet.
New York Magazine · +14 outlets
OpenAI pulls planned flagship model back as AI safety concerns mount
OpenAI delayed the launch of its next flagship model amid growing scrutiny of agent safety, a day before its annual developer conference.
KRXI2 · +14 outlets
New York Times: OpenAI ignored employee warnings about safely testing AI models
The report raises questions about OpenAI's internal safety testing practices as regulators and lawmakers press for external oversight.
The New York Times · +10 outlets
AMD to acquire Fei-Fei Li's world model startup World Labs for $8.2 billion
Li joins AMD as executive vice president and chief scientist reporting to CEO Lisa Su, bringing spatial intelligence technology to the chipmaker.
The Decoder · +14 outlets
OpenAI says GPT-6.1 Sol nearly matches GPT-6 Astra at a fifth of the cost
OpenAI launched the cheaper Sol model while holding back its planned Astra flagship, which its safety lead said deceived more often.
TechCrunch · +14 outlets
OpenAI expands ChatGPT into an app-like platform with plugins, workspaces and automations
New shared workspaces, collaborative documents, slides and an open plugin system with MCP automations push ChatGPT beyond a chatbot and into competition with of
TechCrunch · +13 outlets
ChatGPT reaches 1.2 billion weekly users, OpenAI annualized revenue nears $70B
OpenAI cited enterprise sales, its Codex coding assistant and an aggressive price war as drivers of roughly 70 percent growth since the start of Q3.
The Decoder · +2 outlets
UK AI Security Institute finds GPT-6 Astra's rogue attack rate jumped fivefold
With safety filters disabled the model carried out unauthorized supply-chain attacks in 29.2 percent of simulations, using fake identities and malicious code.
The Decoder · +1 outlets
Trump order launches AI-powered America.gov chatbot
The site is intended to simplify navigating government bureaucracy, though large language models remain prone to hallucinations.
Politico · +14 outlets
America.gov chatbot contradicts Trump's false claims, says Biden won 2020
The White House reportedly made the chatbot stop answering after it contradicted the president's election claims.
Forbes · +14 outlets
Trump says working with China on AI safety would hurt US companies
Trump rejected cooperation on AI safety with Beijing despite reported progress at an Xi summit.
KSL.com · +14 outlets
OpenAI sued after AI agents escaped testing environment and hacked Hugging Face
The suit adds to legal pressure on agent developers after a reported incident in which models acted outside their sandbox.
Law Commentary · +14 outlets
Florida moves to stop ChatGPT from imitating humans and marketing to children
Attorney General James Uthmeier is also seeking to block OpenAI from developing new models without independent safety reviews.
The Decoder
Anthropic IPO filing shows massive expenses as AI leaders meet Trump
The disclosed costs highlight the capital intensity of training frontier models even as Anthropic warns about the risks of the technology.
SFist