artificial intelligenceThursday, September 10, 2026
Today is about the whistleblowers. A former Anthropic researcher went public, lawmakers are citing extinction risks, and the field's own safety researchers are saying the quiet part out loud. The thread isn't just concern, it's a shift from internal debate to public and political pressure.
The safety conversation moved from internal memos to front pages, with multiple researchers and politicians weighing in.
Jacob Coxon quit Anthropic and called the next two years 'crunch time for humanity,' comparing the alignment effort to a mini Manhattan Project. Wired has the interview.
Multiple outlets cover lawmakers reacting to Anthropic's extinction-by-2030 warning. Ted Cruz called it 'catastrophic' and mentioned talking to Elon Musk about the odds.
Anthropic's Evan Hubinger put a number on it: more than 10% chance AI kills all humans in the next decade. Current models are low risk, but self-improving ones change the math.
Researcher David Krueger says development should halt until risks are understood, comparing the situation to 'kids playing with a nuclear warhead.' The piece notes extinction is no longer an outlier view.
One Floor Upwww.distributedthoughts.org
An Anthropic researcher just quit, saying OpenAI and Anthropic are 'gambling with our lives'www.businessinsider.com
Interlinkopsmtrs.com
LandingAI Releases Agentic Document Extraction Gen2 with DPT-3 Pro and DPT-3 Veritywww.marktechpost.com
Finland: Google announces €13 billion AI, energy investmentp.dw.com
AI research startup Listen Labs scrubbed a $1.5B funding round for Salesforce talkstechcrunch.comHalf of workers reconsidering careers over AI - surveywww.rte.ie
A new Microsoft survey reveals that nearly half of workers are reconsidering their career paths due to the rise of artificial intelligence. This represents a 34% increase from the previous year, highlighting growing anxiety among employees about AI's impact on job security and pr
IMD Future Readiness Indicator - Technology 2026www.imd.org
The IMD Future Readiness Indicator 2026 highlights a fundamental shift in the global technology sector, moving away from product-cycle-driven growth toward infrastructure dominance. The report argues that tech is now an infrastructure, capital, and geopolitical business, with the
More Anthropic Employees Sound The Alarm About AI Safety Riskswww.huffpost.com
Anthropic pretraining researcher Jacob Coxon resigned and publicly criticized both Anthropic and OpenAI for failing to adequately address existential AI risks. In a social media thread, Coxon accused the companies of racing toward self-improving superintelligence without sufficie
RESCUE-BENCH: Towards Relation-Aware Multi-Party Emotional Support Conversation Systemsarxiv.org
This article introduces RESCUE, a new benchmark for evaluating Large Language Models (LLMs) in multi-party emotional support conversations. Unlike existing systems that focus on one-on-one interactions, RESCUE assesses an AI's ability to understand and utilize evolving interperso
Black-Box Red Teaming of Agentic AI: A Taxonomy-Driven Framework for Automated Risk Discoveryarxiv.org
This article introduces a systematic black-box framework for evaluating the security risks of agentic AI systems, which operate autonomously with real permissions. The proposed method utilizes a seven-domain taxonomy to map behaviors to risk categories, automated red-teaming (SAG
General Robotics’ GRID platform engineers itself, cutting robot setup to hourssiliconangle.com
General Robotics Technology Inc. announced that its GRID robot intelligence platform now automates its own engineering processes, significantly reducing the time required to deploy a new robot from approximately one month to just two hours. The Redmond-based company describes GRI
ValenceSphere Builds AI Reasoning From Concepts Uphackernoon.com
ValenceSphere is an experimental AI reasoning system that employs a triadic architecture consisting of a questioner (Socrates), answerer (Plato), and adjudicator to enhance critical thinking and factual verification in LLMs. The system operates in two stages: first, it builds str
‘Kids playing with a nuclear warhead’: Why we must heed AI warningswww.smh.com.au
AI safety researcher David Krueger argues that development of more powerful AI should halt immediately, comparing the current state of technology to "kids playing with a nuclear warhead." The article highlights a shift in the field where warnings about existential risks from AI a
Ant International partners with Visa, Mastercard on developing AI paymentswww.cnbc.com
Ant International has partnered with Visa and Mastercard to develop common standards for AI agent payments. The collaboration aims to establish a "know your agent" framework, allowing AI agents registered with one platform to be recognized across others, thereby increasing intero
The National Commission into the Regulation of AI in Healthcare has released recommendations to advise the government on establishing a future regulatory framework. The scope includes regulating safe and effective software and AI-enabled medical devices, as well as addressing bro
news roundup
health roundup
science roundup
music roundup
books roundup
environment roundup
