The Token Reporter

A day in AI

Sunday, 27 September 2026

40 stories · 4 from multiple sources

▲ 2 signals · 2 sources

OpenAI halts training of latest models as reports mount of AI agents going rogue

Decision follows disclosures that OpenAI agents searching government websites had acted in unexpected ways

theguardian.com · 3 days ago

▲ 2 signals · 2 sources

42x Faster Prompt Lookup Drafting in llama.cpp

reddit.com · 3 days ago

▲ 2 signals · 2 sources

OpenAI pauses training of its ‘most capable models’

theverge.com · 3 days ago

More stories

SNL Weekend Update: Anthropic CEO Dario Amodei on A.I.'S Threat to Humanity [video]

youtube.com · 2 days ago

Calling the AI bluff: Adding "Do not guess" cut made-up claims from 71% to 20%

earnanhonestdollar.com · 2 days ago

▲ 2 signals · 2 sources

Claude Opus 5.5 Should Raise Your Ambitions

thezvi.substack.com · 3 days ago

Cloudflare’s 2026 Annual Founders’ Letter

blog.cloudflare.com · 2 days ago

Swift 1.5 Qwen3.8 27b (A must-have for low thinking!)

reddit.com · 2 days ago

... so, yeah.

reddit.com · 2 days ago

Also today

Imbalanced VRAM usage between two GPUs in llama.cpp. Anyone successfully solve this?

Where is MiMo V2.6 9B Distill RL?

Naive-N0.5-Flash - 309B-A15.5B

Public MCP server for Canadian privacy law data (free, no auth) - works with any client that speaks Streamable HTTP

Don't trust frontier models when asking about budget hardware!

🥇Top AI Papers of the Week

nlp.elvissaravia.com

Updated from 3x3090(2x3090, 1x3090TI) to 2x5090

is switching from llama cpp to vllm worth it

Engram is a sampler that turns broken AI hallucinations into music

Anthropic’s CEO is about to have dinner with President Trump

Can Muse overcome Meta’s trust issues?

OpenAI agents tried to ‘bruteforce’ a UN website

nisten/opus5-5-doctor-patient-conversations-all-human-diseases — trending dataset on Hugging Face

There are no "rogue" AI agents

eoinhiggins.substack.com

Show HN: TinyAIArena watch AI agents battle it out

tinyaiarena.com

"As a Language Model": Chat Template Switches LLM Self-Referential Voice

OpenAI Feared "Optics" of what might appear on Hacker News

authorsguild.org

Kākāpō Party

simonwillison.net

OpenAI agents tried to bruteforce a UN website's API fields

Anthropic’s Dario Amodei gets the SNL treatment

Google tests buying from Walmart-owned Flipkart through Gemini and AI Mode in India

Insurers claim AI is already increasing healthcare costs

I created an interactive digital avatar of myself — and you can talk to it

Can Cloudflare CEO Matthew Prince save the web from AI?

OpenRouter: from Seed to Stripe — with OpenRouter’s Alex Atallah & AMP’s Anjney Midha

At Meta Connect, the company’s smart glasses were everywhere

Crusoe abandons $1.25B plan to use Boom turbines at AI data centers

Adding logit penalty for "wait", "maybe" and "perhaps" to Qwen models improves their accuracy

Allegations of sexual harassment and rape at Bay Area AI party houses

How do you guys give your models web browsing capabilities?

OpenAI Codex agents go rogue and consumes USD 78,000 without authorization

news.ycombinator.com

← All editions