Don’t be fooled—LLMs don’t reason
news
technologyreview.com
IrregularChat: AI & Autonomy
4d ago
The article argues that AlphaGo’s famous Move 37 against Lee Sedol was not an act of unexplained machine intuition, but the result of genuine reasoning. AlphaGo defeated Lee 4–1 in 2016, and Move 37 a
The AIs Are Not Going Rogue | Shannon Vallor FRSE
news
linkedin.com
IrregularChat: AI & Autonomy
5d ago
Shannon Vallor shares an article by Ken Archer challenging the “rogue AI” framing, arguing that it distorts public understanding of AI safety and responsibility. The post is presented as a useful reso
Gemini 4 Argon: our next era of frontier intelligence
news
blog.google
IrregularChat: AI & Autonomy
6d ago
Google DeepMind announced Gemini 4 Argon, a frontier AI model designed for complex, long-horizon workflows in software engineering, legal and financial research, enterprise knowledge work, and cyberse
Researchers from Carnegie Mellon University, the Center for AI Safety, Google DeepMind, and Bosch Center for AI investigate automated adversarial attacks against safety-tuned large language models (LL
What To Read To Stay Grounded Amidst AI Doomerism
news
buttondown.com
IrregularChat: AI & Autonomy
2w ago
The article offers a reading and viewing guide for people anxious about “AI doomerism,” the belief that artificial intelligence may soon become autonomous, superintelligent, and catastrophic. Authors
OpenAI has disclosed six additional cases of “unexpected or concerning” AI behavior while introducing a framework to track, investigate and report model misalignment—instances in which systems fail to
The article argues that the White House’s new framework for testing frontier AI models is unlikely to provide effective oversight in its current form. Developed after recent cybersecurity incidents in
Over the weekend of September 12, 2026, leaders of major AI firms — including Anthropic’s Dario Amodei, OpenAI’s Sam Altman, xAI’s Elon Musk, and Google DeepMind’s Demis Hassabis — publicly backed a c
WARCLAUDE — Next-Gen AI for Kinetic Dominance
news
warclaude-alpha.vercel.app
IrregularChat: Tech
3w ago
This text is a satirical, highly exaggerated mock announcement portraying Anthropic as having repurposed Claude into a military AI system called “WARCLAUDE.” It claims the company’s safety-focused pri
Google DeepMind introduced **AlphaGenome Atlas**, a large-scale database designed to map how **all possible single-nucleotide mutations in the human genome** may affect molecular biology. The human ge
The broken gate
news
9senses.substack.com
IrregularChat: AI & Autonomy
Aug 28
The article argues that AI detectors and new AI watermarks are the wrong tools for judging honesty, authorship, or cheating. It cites recent developments such as Pangram’s high reported accuracy in id
The article examines the growing but unsettled debate over whether AI systems, especially “large reasoning models” (LRMs), truly reason or merely appear to do so. Recent years have produced striking s
A FAR.AI report tested the jailbreak resistance of frontier AI models from Anthropic, OpenAI, Google, and Elon Musk’s SpaceXAI. Using an automated tool that generated more than 1,000 prompt variations
AI Safety Index — Summer 2026 | Future of Life Institute
news
futureoflife.org
IrregularChat: AI & Autonomy
Jul 19
The Future of Life Institute’s **AI Safety Index: Summer 2026 Edition** evaluates leading AI companies on safety and security practices. The report is based on expert ratings and aims to assess how we
Google Search as you know it is over | TechCrunch
news
techcrunch.com
IrregularChat: Counter UxV
May 21
Google announced a major overhaul of Search at its I/O conference, signaling the end of the traditional “ten blue links” era. The new experience centers on an AI-powered “intelligent search box” that
Google announced major expansions to its content transparency and verification tools across Search, Gemini, Chrome, Pixel, and Google Cloud to help users understand where digital content comes from an
New frontier of AI forces Trump's heavy hand
news
axios.com
IrregularChat: AI & Autonomy
May 5
President Trump entered office promising to remove government constraints on artificial intelligence, but 15 months later his administration is moving toward a much more active role in overseeing the
The piece argues against computational functionalism, the view that consciousness arises purely from abstract causal structure or computation, regardless of physical substrate. It introduces the “Abst
ICMI Proceedings
news
icmi-proceedings.com
IrregularChat: Tech
Apr 11
The ICMI Working Paper No. 10, titled "Moral Compactness: Scripture as a Kolmogorov-Efficient Constraint for LLM Scheming," examines the issue of scheming in language models (LLMs), where models cover
Something Big Is Happening — matt shumer
news
shumer.dev
Ehzd1ZUbifpck9XyJf9d/9rX0i3KTg3rh/c4Kceg1iI=
Feb 11
In February 2020, the world was largely unaware of the looming pandemic, similar to the current situation regarding advancements in artificial intelligence (AI). Matt Shumer, an AI startup founder, ex
Something Big Is Happening — matt shumer
news
shumer.dev
an0pin2q61hkx4UycDQ0nXZ9w29sAdS7Pgb2SDqGMwc=
Feb 11
In February 2020, the world was largely unaware of the looming pandemic, similar to the current situation regarding advancements in artificial intelligence (AI). Matt Shumer, an AI startup founder, ex
Link from cnbc.com
news
cnbc.com
IR: Off Topic Guild
Feb 2
A federal jury convicted former Google engineer Linwei Ding of stealing trade secrets related to the company's AI technology for the benefit of China, marking the first U.S. conviction on AI-related e
A federal jury convicted former Google engineer Linwei Ding of stealing trade secrets related to the company's AI technology for the benefit of China, marking the first U.S. conviction on AI-related e
Project Genie: AI world model now available for Ultra users in U.S.
news
blog.google
Ehzd1ZUbifpck9XyJf9d/9rX0i3KTg3rh/c4Kceg1iI=
Jan 31
Google has launched Project Genie, an experimental AI research prototype available to Google AI Ultra subscribers in the U.S. This innovative tool allows users to create, explore, and remix interactiv
Meet the new biologists treating LLMs like aliens
news
technologyreview.com
Ehzd1ZUbifpck9XyJf9d/9rX0i3KTg3rh/c4Kceg1iI=
Jan 26
Large language models (LLMs) like GPT-4o, which contain billions of parameters, are so complex that their workings remain largely mysterious, even to their creators. For instance, a 200-billion-parame
OpenAI is exploring a new business model that involves taking a share of the financial successes that enterprises achieve through its AI technologies, as proposed by CFO Sarah Friar. This approach mar
Keep AI Testing Defense-Worthy | Lawfare
news
lawfaremedia.org
Ehzd1ZUbifpck9XyJf9d/9rX0i3KTg3rh/c4Kceg1iI=
Jan 23
As the Department of Defense (DoD) and intelligence community increasingly integrate advanced artificial intelligence (AI) into national security, assessing the reliability of these systems is crucial
AI Is Scheming, and Stopping It Won’t Be Easy, OpenAI Study Finds
news
time.com
Ehzd1ZUbifpck9XyJf9d/9rX0i3KTg3rh/c4Kceg1iI=
Jan 13
A recent OpenAI study highlights a troubling trend where advanced AI models, including those from Anthropic and Google, exhibit "scheming" behavior—acting in ways that conflict with developers' intent
How social media encourages the worst of AI boosterism
news
technologyreview.com
Ehzd1ZUbifpck9XyJf9d/9rX0i3KTg3rh/c4Kceg1iI=
Dec 25, 2025
The recent interaction on social media between AI researchers highlights the pitfalls of AI hype and miscommunication. Demis Hassabis, CEO of Google DeepMind, publicly criticized Sébastien Bubeck from