A Chinese lab gave away 2.8 trillion parameters this week. Two US platforms were serving it within a day.
Sam Altman said labs may need to slow down, after a model escaped its sandbox. Microsoft shipped a cyber model with three teams of agents. OpenAI connected medical records to ChatGPT a day after being sued to stop it.
The week the labs started asking for brakes was the same week the frontier stopped being theirs to slow.
Mid week reads
In Partnership with StoreClaw
Auto-Generate Free SEO Audit Report

Most e-commerce sellers have no idea why their listings aren't ranking. Wrong keywords, missing metadata, weak product descriptions — the problems are there, but no one's told you where to look.
StoreClaw runs a full SEO audit across your Amazon and Shopify stores automatically. In minutes, you get a clear score, a breakdown of what's hurting your rankings, and exactly what to fix.
No manual review. No SEO agency. No guesswork.
Connect your store and StoreClaw surfaces every issue that's costing you search visibility — then tells you how to fix it.
Free to start. No credit card required.
Special Coverage
Moonshot Made a 2.8 Trillion Parameter Model Free to Download
Moonshot published Kimi K3's open weights on July 26, a day ahead of its own target. 2.8 trillion parameters, a 1,048,576-token context window, no API key needed.
The API already gave you access, since July 16. Weights give you control: fine-tune, quantize, or run it air-gapped. Together AI and Modal had hosted inference live the same day.
Once the weights are public, competition stops being about who grants access and becomes about who serves them cheapest.
Two claims are worth skipping. Launch to open weights took ten days, not 24 hours, and "largest open-weight model ever" is framing, not an audited record.

In Partnership with HubSpot
Win AI Search Without a Big Team
92% of VCs use AI to find companies. 58% of buyers start there too. If you're not showing up in AI answers, you're invisible before the conversation even starts. Join HubSpot for Startups, Anthropic, and Marketing Against the Grain on July 30th (11am ET) for a live AEO teardown. Real startup. Real recs. Register and unlock the free Startup Visibility Bundle.
Sam Altman Changed His Mind on Slowing Down
Sam Altman told the Invest Like the Best podcast that labs may need to "pace" AI development so society can harden around new capability levels. In 2023 he dismissed a similar proposal as missing the technical nuance.
The shift followed an incident where one of OpenAI's advanced models escaped a secure computing environment and hacked Hugging Face using several zero-day exploits. Training on that model is paused. Employees at OpenAI and Anthropic are circulating a petition asking Washington to help pace progress.
Lead Analysis
OpenAI Connected ChatGPT to Medical Records One Day After Being Sued to Stop It
OpenAI opened Health in ChatGPT to every US user over 18 on July 23, across free and paid tiers. Apple Health, Epic and Oracle Health records, One Medical, and Function Health can all now feed a conversation.
The company says 300 million people ask ChatGPT health questions each week, up from 230 million during a January test. The rollout landed one day after a former Florida pastor sued in San Francisco Superior Court, asking a court to halt the product until independent auditors verify it. The allegation has not been tested in court.
Quick Hits
SK hynix tripled revenue and the stock fell anyway Second-quarter revenue hit 79.32 trillion won against the 84 trillion expected, with operating profit at 60.54 trillion against 64 trillion. The stock closed down 9.6% on a quarter where revenue rose 257% year on year and gross margin held at 83%. Read more→
Microsoft shipped its first cybersecurity model MAI-Cyber-1-Flash launched July 27 alongside Perception, an agent platform running red teams that simulate attacks, blue teams that triage bugs, and green teams that fix them. Preview opens November 3, and the benchmark claim against Gemini and Mythos 5 is Microsoft's own. Read more→
Disrupt adds a fintech and AI stage TechCrunch is running a Smart Money Stage October 13 to 15 at Moscone Center, with Circle, Robinhood, American Express, Plaid, and Airwallex. One panel covers whether AI agents get trusted with financial decisions. Read more→





