AI

LLMs aren't conscious (and thinking they are is culturally dangerous) (theintrinsicperspective.com) AI

The article argues that large language models like ChatGPT are not conscious, and that claiming they are can be culturally harmful, though it acknowledges that their capabilities can still be genuinely impressive without requiring consciousness.

Amazon and the White House Ended Anthropic's Fable (axios.com) AI

The article argues that Amazon and the White House effectively ended Anthropic’s “Fable” effort, based on developments referenced by the story’s headline and URL.

Extinction-level capitalism (matthewbutterick.com) AI

The essay argues that AI is inherently political and, even without malfunctions or malicious actors, could erode liberal democracy by amplifying existing trends like capital concentration, leading to an irreversible shift in political and economic arrangements.

The Curse of Depth in Large Language Models (arxiv.org) AI

The arXiv preprint “The Curse of Depth in Large Language Models” argues that increasing model depth can create negative effects for large language models, based on evidence and analysis described in the paper.

AI forgoes toxic positivity for neurodivergents (medium.com) AI

The piece argues that conventional productivity and planning tools fail many ADHD and autistic adults by demanding rigid routines that worsen “waiting mode,” guilt, and shame. It says neuro-affirming design should instead offer low-demand, judgment-free support—citing an experimental conversational AI called Neuro+ as a real-time cognitive companion.

PwC Report: AI Making Medical Bills Higher (fortune.com) AI

A new 60-page PwC report says hospitals’ use of AI—particularly for note-taking and coding—has been helping drive higher medical bills by enabling more granular, higher-severity diagnosis codes, even when patient care is unchanged.

US ban on Mythos is related to a jailbreak research by Amazon researchers (timesofindia.indiatimes.com) AI

The US ordered Anthropic to suspend access to its Fable 5 and Mythos 5 AI models over national security concerns, and the report says the issue traces back to a jailbreak approach tested by Amazon researchers using prompts to induce the models to reveal security vulnerabilities.

Amazon CEO's Talks with U.S. Officials Triggered Crackdown on Anthropic Models (wsj.com) AI

The article says discussions between Amazon’s CEO and U.S. officials reportedly led to a U.S. crackdown on Anthropic models, based on the claim reflected in the headline since no article text was provided.

AI Coding at Home Without Going Broke (stephen.bochinski.dev) AI

The article outlines three ways to do AI coding at home affordably—self-hosting open-source models, renting models via API providers like OpenRouter, or using reduced-cost “min-maxed” subscriptions from OpenAI and Anthropic—and argues that a hybrid approach (frontier models for planning/specs, cheaper open-source models for routine tasks) can deliver team-level output for relatively low monthly costs.

GLM 5.2 Is Out (digg.com) AI

Z.ai has launched GLM-5.2, a coding-focused model advertised with a 1-million-token context window and immediate availability for subscribers to its GLM Coding Plan. The company says API and chatbot services will roll out next week and that the model will be officially open-sourced under the MIT License next week, though the exact open-source timing remains unclear.

repo-slopscore: Detecting AI/LLM contributions in git repositories via commit history analysis (slopscan.ava.pet) AI

repo-slopscore is a tool/service that scans git repositories using commit-history analysis to detect or estimate potential AI/LLM contributions.

Talk more to your coding agents (datawill.io) AI

A June 5, 2026 post argues that the most effective coding-agent workflow is to spend more time talking with the agent to clarify assumptions and questions, then have the agent distill that discussion into a plan and implement via separate issues/PRs, followed by a final review and an extra “polish” pass.

Dangerous Technology for Americans Only (lucumr.pocoo.org) AI

In a June 13, 2026 blog post, Armin Ronacher argues that U.S. export controls on Anthropic models—framing them as “dangerous” and restricting access to foreign nationals—reflect a shift from safety to nationalism, effectively treating frontier AI like a weapon. He warns this approach could deepen global division and leave Europe dependent on U.S. power, criticizing European reliance on regulation over strategic capability and calling instead for stronger European capacity and more international cooperation, including open source.

Claude Fable 5 vs. GPT-5.5: Better Planning, Similar Execution (blog.kilo.ai) AI

Kilo.ai reports a comparison of Anthropic’s Claude Fable 5 and OpenAI’s GPT-5.5 for building a feature-flag rollout service: Claude generated the better development plan, but when both models implemented the same winning plan, they produced identical behavior and passed all 15 acceptance checks, with GPT-5.5 costing substantially less; the post attributes the planning gap to sharper handling of edge cases and clearer commitment to contested design choices (e.g., caching misses and key hashing), while execution quality was effectively the same.

Google proposes Open Knowledge Format based on Markdown (cloud.google.com) AI

Google Cloud announced the Open Knowledge Format (OKF), an open, vendor-neutral specification for packaging AI “LLM-wiki” knowledge as a directory of Markdown files with YAML frontmatter. OKF v0.1 is designed so knowledge producers and AI agents can exchange context, metadata, and curated documentation without proprietary SDKs, with optional index.md and log.md to support navigation and change history. The post also describes reference implementations (a BigQuery-to-OKF enrichment agent and an HTML visualizer), sample OKF bundles, and plans to evolve the specification via GitHub contributions.

GLM-5.2 is now available with 1M-context support (twitter.com) AI

A social media post says GLM-5.2 is now available with 1M-context support, according to the link title, with no further article details provided.

Agent Memory Systems and Knowledge Graphs: Letta, Mem0, Graphiti, and Cognee (codepointer.substack.com) AI

The piece compares four long-term agent memory/knowledge-graph approaches—Letta, Mem0, Graphiti, and Cognee—showing how each handles storing, retrieving, and updating facts across sessions, from simple self-edited text blocks (Letta) to entity-linked vector memories (Mem0), bi-temporal typed fact edges (Graphiti), and an explicit extract-cognify-load pipeline that builds graph and vector stores (Cognee).

OpenAI under investigation by group of state attorneys general, source says (reuters.com) AI

OpenAI is reportedly under investigation by a coalition of state attorneys general, according to a source cited by the WSJ, amid claims detailed in a Reuters report.

GLM 5.2 Released (twitter.com) AI

The post announces the release of “GLM 5.2,” but provides no additional details because no article text was available.

RTX 5080 and RTX 3090 Setup: 80 Tok/s on Qwen 3.6 27B Q8 (imil.net) AI

The post describes a local multi-GPU Linux setup pairing an RTX 5080 with an RTX 3090 to run Qwen 3.6 27B quantized to Q8, focusing on BIOS/PCIe settings, driver considerations, and llama.cpp build/run parameters; the author reports throughput around 80–90 tokens per second with speculative decoding enabled (draft MTP/“ngram-mod”).

Cerebras chips rival Nvidia GPUs for AI [video] (youtube.com) AI

The video discusses Cerebras AI chips as potential rivals to Nvidia GPUs, based on the article title and link information provided.

AI OSS tool repo goes archived over night after raising $7.3M Seed (github.com) AI

The TensorZero GitHub repository (an open-source LLMOps platform for unifying an LLM gateway, observability, evaluation, optimization, and experimentation) was archived by its owner on Jun 12, 2026, making it read-only. The repo’s README states the project raised a $7.3M seed round and describes features such as a low-latency Rust gateway, evaluation tooling, and A/B testing/routing, though the provided data does not explain why it was archived.

Thoughts on AI and Jobs (blog.keyvan.net) AI

In “Thoughts on AI and Jobs,” Keyvan argues that while people deserve sympathy as AI threatens livelihoods, jobs themselves are not inherently worthy of protection and can be soul-crushing or undemocratic; he says AI already performs much of the repetitive work in both manual and “knowledge” roles and will likely drive major changes to how people work, even if it cannot replicate human-level intelligence.

I'm open-sourcing Bottega, our internal coding agent orchestration tool (vdaubry.github.io) AI

The post announces the open-sourcing of Bottega, an internal tool that orchestrates coding agents in a human-in-the-loop workflow where a developer reviews a detailed plan and agents handle implementation, code review, testing, and PR/CI management; the author says this approach reduced PR back-and-forth and improved quality, and the tool is now extended to support multiple model providers (Claude Code plus Codex/OpenCode) with model choice per pipeline step.

KPMG's AI report turns into a demo of AI hallucinations (theregister.com) AI

A review by GPTZero claims that KPMG’s October 2025 AI report, “Total Experience: Redefining Excellence in the Age of Agentic AI,” contains widespread citation problems—only 5 of 45 citations reportedly match their sources—and that many factual claims, including a case study about Emirates’ “Sara,” appear to be false or unreliable. The report has since been removed from some KPMG websites while the firm investigates how it was published, with KPMG saying it is reviewing circumstances and emphasizing human oversight to validate content and sources.