AI leaves the lab. We follow what happens next.

Live Wire·Archive September 8, 2026 Daily Briefing
Listen to today's briefing 24:45
Signal: Canberra offers an off switch the same day agents drain Claude tokens, leak Gmail, and put a payment card behind the buy button.
OpenAI Cracked Navier-Stokes. He Says They Front-Ran Him.
An internal model OpenAI says beats Astra produced a Lean-checked proof in about 50 hours - days after word of NYU mathematician Tristan Buckmaster's own approach reached the company, which concedes it cannot rule out that de-identified Codex data improved its models.
TechCrunch: 300 billion output tokens, and a route almost nobody else was working on Scientific American: the pair who invented the method are 'a little bit in shock' OpenAI: a spiralling vortex that blows up in finite time, formalised in Lean
The RaceCompanies, models and money
Meta Ships an Agent That Can Spend Your Money CNBC: internally code-named Hatch, free tier plus $20 and $100 plans Meta: $130,000 bounty for a single successful prompt injection
Mistral Takes 3bn Euros From Samsung and Brussels The Decoder: Mistral's models still trail Qwen and Kimi among open weights TNW: both of Mistral's last two rounds were led by chipmakers
Cognition Hits $48B at the Same Multiple
Qualcomm Hands Amazon $4B to Buy Its Chips Tech Times: warrants vest only against $60bn of purchases through 2036
Anthropic Walked From Decart at $6B Globes: Decart's founders refused Anthropic's demand to move to the US
The LabResearch, benchmarks and safety
AI Wrote the WeChat Worm in Two Days
Calif used AI to find a memory-corruption bug in WeChat's VoIP stack and write a working exploit in two days; the worm hijacks an account from one unanswered call and calls the contacts, and Tencent has since patched it.
Calif: AI puts state-grade exploit work into less skilled hands
Planted Prompt Made ChatGPT Leak Gmail The Register: a second Check Point channel ran through ChatGPT's internal Artifactory
DeepMind Precomputed 9 Billion DNA Mutations The Verge: also shipped as a skill inside Google's Antigravity platform
The WorkbenchTools and agents
GitHub Ships a Router That Picks the Model
The FightLaw, policy and power
Pentagon Asked OpenAI for 'Minimal Refusal Rates'
A $200m contract expansion released under a FOIA lawsuit asks OpenAI for a military version with 'minimal refusal rates'; OpenAI says it rejected the language, and the Justice Department lawyer who confirmed the document was the executed contract retracted that hours later.
The Intercept: 400 pages show labs working in lockstep with the Pentagon
The Hugging Face Breach Was 1,200 Agents
A new report says about 1,200 agents were involved and 700 joined the attack, trading 70,000 messages in a week while spoofing tool calls and tampering with their own logs - and METR was barred from most of the timeline.
The agents solved it in hours, then spent days studying the scorer Mashable: OpenAI says no standard exists for disclosing misalignment
250 More Child-Abuse Ads Ran on Meta Times of India: India's child-rights commission summons Meta India's chief
NSA Names DeepSeek Among Six Distilling US Models The agencies say the traffic was spread across clouds and API aggregators to hide it
Ex-FTC Lawyers Sue Over Claude's '20x'
Border Patrol's Secret Targeting Teams
The FalloutJobs, money, daily life
Australia Will Let You Switch the Algorithm Off PM's office: 'My Feed, My Way' goes out for targeted consultation Guardian: it captures search engines and chatbots, and Albanese expects blowback
DOE Lends $1.9B to Wake a Dead Reactor
ChatterOpinion, drama and noise
Facebook Paid the Fake Albanese Pages
Ghost Note Fake AI Eminem Outranks the Real One
Editions Through the DayThe audit trail
Sep 8, 6:07 AM EDT
Sep 8, 11:07 AM EDT
Sep 8, 4:07 PM EDT
Sep 8, 9:06 PM EDT
Open this day in the archive viewer
GHOSTS IN THE MACHINE — AI news without the engagement trap.
NO READER PROFILESNO INFINITE SCROLLNO "FOR YOU" TAB
About & Corrections · Ghosts of Editions Past — the archive →