OpenAI announces three big science discoveries and a new way to work with academics
Less than a month after its mass release of hundreds of manuscripts that it said contained solutions to long-standing mathematics…
AI · Tools · Online creation
Latest AI news
Every AI story worth reading, collected hourly from the labs, the tech press, research and the creator economy. Sorted by time, grouped by topic.
3,603 stories · Every headline links to the original publisher. HapnGo curates and groups; we do not copy articles.
PulseDev
Click any UI bug on localhost and your AI agent fixes it Discussion | Link
Cross-request KV cache exposure in the GLM-5.3 Workers AI model (155 responses)
AI summary Cloudflare said a flaw in its Workers AI deployment of GLM-5.3 let 155 responses include data from other customers' recent requests between August 27 and September 9, 2026. The leak, reported on September 8 by Gadi Evron of Knostic, hit only prompts exceeding 681,280 tokens, and the company found no sign of deliberate exploitation. It stemmed from a KV cache pool smaller than the 1,048,576-token context window, combined with a bug in the open-source SGLang engine. After fixing both, Cloudflare restored the model on September 9 and urged operators to move to SGLang v0.5.22 or later.
Do you feel more anxiety the more you AI tools?
The title - Do you feel more anxiety the more you AI tools? My observation is that, I'm heaps more productive with the help of AI…
Google AI Constant Errors
I used the regular Google search for answers about my new Toyota Sienna. I find that if you use a full question, it gives an AI…
In a rush to lure AI companies, the Australian government risks undoing a creative ecosystem that has taken decades to build
If proposed copyright reforms proceed, Australia will become a global IP-laundering paradise. Why would a UK business license its…
Codex of Weights: an open ethical protocol for AI built on mutual obligations, not just control
Most AI safety discussion is about how to control machines. I've been working on an open document that asks a different question…
Greptile takes bug detection beyond code review with GPT-6 Astra
How Greptile uses GPT-6 Astra to test user workflows, Daybreak to strengthen security, and Codex to reduce rework.
Anthropic is burning gas to feed the AI boom
As it prepares to go public, Anthropic's S-1 filing makes generative AI's methane dependency glaringly obvious. "Competition for…
Revix AI
AI code review that learns your repo's own rules Discussion | Link
Why Anthropic’s IPO Prospectus Is Missing So Many Key Figures
Anthropic’s S-1 filing is hundreds of pages long—and it leaves out quite a few financial details. “Anthropic owes investors more…
Google Ads Expands Customer Lifecycle Goals In Search And Shopping
AI summary Google Ads will begin rolling out new customer lifecycle features in October, Ads Liaison Ginny Marvin announced on LinkedIn. High Value mode, which lets advertisers bid more for new buyers resembling their best existing customers, is moving to Shopping campaigns after launching in Search and Performance Max. After beta testing, the Loyalty, Re-engagement and Retention-only modes are becoming broadly available in Shopping, Search and Performance Max. High Value requires value-based bidding and a minimum of one Purchase conversion goal.
Has AI changed how you look things up online, or do you still mostly just use Google?
I notice I'm reaching for a chatbot more often for quick questions, but for some stuff I still go straight to a search engine…
Google Updates Spam Guide For Shared Pages, Generated Content & Apps
AI summary On Oct. 26, Google revised its spam guidance for hosting platforms, adding a section telling services that let users make shareable pages, interactive tools or generated content to keep them out of search by default with noindex or a robots.txt block. The guide, previously updated on Dec. 10, 2025, now also recommends verifying sites in Search Console, watching indexing and performance reports, and setting alerts when a few users produce a large portion of indexable URLs. Google's own documentation notes that a robots.txt block stops Googlebot from reading a noindex rule, and blocked URLs may still appear in the index when other sites link to them.
GPT-6.1 Terra halves OpenAI's API prices and still gains ground on its larger models
AI summary OpenAI has launched GPT-6.1 Terra, a mid-range model that costs $1 per million input tokens and $5 per million output tokens, half the price of GPT-5.6 Terra. It is offered through the API and in ChatGPT Work and Codex for paid subscribers. On Terminal-Bench 4.0 it solved 50.3% of tasks against 21.5% for its predecessor, and on AA-Omniscience it guessed 38% of the time when it lacked an answer, down from 88%. For the toughest coding jobs, OpenAI still points users to GPT-6 Astra and GPT-6.1 Sol.
Proton To ‘Re-Interview’ Staff About AI To See Who Gets To Keep Their Job
“These re-interviews are consequential,” a leaked email seen by 404 Media reads.
Nuvio 5.0
Hand your financial admin to your AI agent Discussion | Link
Apple releases iOS and macOS 27.2 with new Siri AI features, new Music logo
Included is the ability to limit which applications Siri can directly access.
AI assistant traffic to retail sites doubled this year. Those visitors convert at half the rate of organic search.
An analytics firm tracked 2,000 retail sites and found both things are true at once: visits originating from AI assistants…
Deepity: a hand-tuned C++/CUDA predictive coding library, beats jpc (JAX) and ngc-learn on MNIST (CPU benchmarks + source inside) [P]
I've been building a predictive coding (PC) engine from scratch in C++20, with a thin Python layer on top, and just cut the first…
throxy
Cold calls in your prospect's language. Pipeline delivered. Discussion | Link
Google Warns Tools On User-Generated Spam Guidance & More
AI summary Google has revised its help page on fighting user-generated spam so it now also covers services where users publish shareable pages, generated output or web apps. A new section advises such platforms to keep link-shared, collaborative or utility pages out of the index by default, using a noindex rule or robots.txt blocking. Operators are also told to verify their sites in Search Console and watch for sudden jumps in indexed pages or impressions. On X, Marie Haynes asked whether manual actions against AI and app-building tools may follow, while Lily Ray pointed to Claude artifacts and ChatGPT pages.
Anthropic opens its books for Wall Street, showing record revenue and mounting obligations
AI summary Anthropic has submitted an S-1 prospectus to the SEC to list on the Nasdaq as "ANT", in an offering led by Goldman Sachs and Morgan Stanley. The filing reports $11.6 billion in revenue for the second quarter of 2026 and a first GAAP operating profit of $974 million. Against that, the company has committed roughly $780 to $810 billion to long-term compute deals with Amazon, Google, Broadcom and others. Token-based usage billing makes up 80 percent of revenue, and Wall Street reportedly expects a valuation near $2 trillion.
Free, daily
The HapnGo Briefing, in your inbox
One email a day with the AI stories that matter, why they matter, and one tool worth trying. No spam, one-click unsubscribe.