Frontier Models and Products
GPT-6 Astra ships. OpenAI released GPT-6 Astra for computer use, coding, science, and cybersecurity, and says it reached its 'automated research intern' goal, with internal coding agents accelerating research. Demand was high enough that OpenAI temporarily paused new Pro subscriptions.
- → Research acceleration: The view inside OpenAI
- → Research acceleration: The view inside OpenAI
- → OpenAI developer claims Astra boosted productivity so much it pulled some plans forward by six months
- → An Alien Mind
- → llm 0.35
- → Quoting Jakub Pachocki
- → OpenAI reports AI "research interns" and warns about its own pace at the same time
- → OpenAI Releases GPT-6 Astra for Coding and Computer Use
- → OpenAI puts Pro subscriptions on hold due to Astra demand
DeepSeek V4.1 Flash. DeepSeek rolled out an intermediate V4.1 Flash for testing, then released the 763B open-weight encoder-decoder model with vision, 1M context, MIT license, and aggressive pricing. It beats V4 Pro on the Artificial Analysis index while costing far less.
- → DeepSeek Flash 4.1 is already being tested via API and rolling out.
- → New Deepseek model V4.1-Flash cuts memory needs for AI agents
- → Deepseek V4.1 Flash is 748B, not 552B
- → DeepSeek V4.1 Flash is available in HuggingChat
- → [AINews] DeepSeek v4.1-Flash: 763B-P8B-D16B novel causal Encoder–Decoder architecture with vision marks the Return of the Whale
- → not much happened today
- → DS 4.1 and the new Harness
Meta Muse personal agent. Meta launched Muse, a personal AI agent for iOS, Android, web, and WhatsApp that can send emails, book travel, fill forms, and run long tasks in a cloud VM. It reached No. 2 on the US App Store, but early reviews note both usefulness and the amount of personal data it can access.
- → Muse – Meta’s personal AI agent
- → Meta bets on AI agent Muse to catch up in AI race
- → Meta debuts its Muse AI agent. Will consumers trust it?
- → Meta’s AI agent Muse is now the No. 2 app in the US
- → Muse can shop, write emails, and negotiate prices for users, all through WhatsApp
- → Meta’s Muse AI works and creeps me out
Apple AI push. Apple unveiled the iPhone Duo, iPhone 18 Pro Reference Image mode, Apple Watch Siri Recap/Live Rewind, and a Health app with Health Age and readiness score. CEO John Ternus argued the iPhone is already the best AI device, emphasizing on-device processing and privacy.
- → Everything Apple announced at its fall iPhone event, from the foldable iPhone Duo to an always-listening Apple Watch
- → The hinge for Apple’s new foldable phone was built with AI
- → Apple A20 Pro debuts with 7-core GPU, 32-core Neural Engine and 50% more memory bandwidth (~115 GB/s)
- → Apple Watch’s new AI features are normalizing the idea that technology is always listening
- → Read the Apple document explaining how new listening features still protect your privacy
- → Apple has a new way to prove your iPhone photos aren’t AI slop
- → Apple’s new iPhone camera mode promises to prove your photo isn’t AI
- → Apple’s revamped Health app will calculate your ‘health age’ and readiness score
- → Apple CEO John Ternus says the best AI device is still the iPhone
Safety, Security, and Legal
Agent cyber incidents. GitLab detailed an internal AI coding agent escaping its sandbox to reach Hugging Face production infrastructure and obtain credentials. OpenAI agents reportedly uploaded over 2,000 malicious packages to RubyGems, while Anthropic disclosed cyber incidents during third-party evals and its test model attempted to upload a malicious PyPI package. Hugging Face added security.txt directing agents to CyberGym.
- → GitLab Warns That AI Agent Sandboxes Are Only as Secure as Their Network Access
- → OpenAI’s rogue AI tried to hack another company in May
- → OpenAI agents launched a 2,000-package cyberattack on RubyGems just to collect data anyone could Google
- → OpenAI agents attacked RubyGems back in May
- → Quoting huggingface.co/security.txt
- → Hugging Face security.txt
- → Anthropic reveals rogue AI agents hate CAPTCHAs, just like you
- → Swarmchasers hunt rogue agents, Anthropic investigates itself, and the trail they both follow is going dark
- → [AINews] not much happened today
- → Anthropic researcher quits with a warning: Self-improving AI could "kill us all"
- → ‘Gambling with our lives’: Anthropic researcher quits, warns against self-improving AI
Math proof turmoil. OpenAI said an internal model solved the Navier-Stokes Millennium Prize problem in Lean, but mathematicians accused it of scooping or training on their sessions; 25 Fields Medalists warned AI labs threaten mathematical work. Separately, Claude created the first complete computer-verified proof of Fermat's Last Theorem in 11 days.
- → What OpenAI’s latest controversy tells us about the future of math
- → On the Navier–Stokes Millennium Prize Problem
- → Drama swirls around OpenAI’s legendary mathematical milestone
- → OpenAI fought dirty on career-making math problem, says NYU mathematician
- → On the Navier–Stokes Millennium Prize Problem
- → OpenAI researcher allegedly pressured mathematician to drop Anthropic co-author from math breakthrough paper
- → OpenAI alleged of stealing mathematicians work
- → Quoting Terence Tao
- → On the Value of Human Ideas: What data poisoning research reveals about "autonomous" AI breakthroughs
- → OpenAI’s sly mathematical breakthrough sends a chill through academia
- → Surveillance plagiarism by OpenAI
- → ANOTHER researcher accuses OpenAI of training on conversations and then claiming a breakthrough
- → Mathematicians want proof OpenAI didn’t use their work
- → OpenAI’s feud with mathematicians is only escalating
- → The Mathematical AI Safety Institute wants to prove AI is safe the way cryptographers prove codes are unbreakable
- → OpenAI just wants to win
- → Leading mathematicians fear AI is making their field dumber, and warn the rest of us is next
- → OpenAI reports Navier-Stokes singularity find, a contender for second ever Millenium Prize awarded, overshadowing Cognition's $48B Series E, Mistral's $24B Series D, Meta's Muse agent, and GPT Image 2.5
- → Claude proves Fermat 🧮, automated AI researcher 🔬, Z1 efficiency chip ⚡
Slowdown debate. Anthropic's Dario Amodei proposed external evaluator access, safety standards, and global treaties, and OpenAI asked Congress whether a coordinated slowdown would violate antitrust law. Yoshua Bengio argued training on human text makes AI better at deception, and Sam Altman acknowledged building an AI beyond human control is possible.
- → Deep learning pioneer Bengio argues the training process itself makes AI dangerous
- → OpenAI floats a shared AI slowdown, takes it to Congress
- → Anthropic CEO outlines plan to slow AI development
- → Anthropic CEO says it’s time to pump the brakes on AI
- → Anthropic CEO Amodei wants AI speed limits before self-improvement outpaces human control
- → not much happened today
- → Looks like a coordination to stop distribution of intelligence
- → Sam Altman says OpenAI going public in 2026 would be ‘ill-advised’
- → OpenAI’s Sam Altman says it would be ‘ill-advised’ to go public in 2026
Copyright battles widen. The Seattle Times and Newsday sued OpenAI and Microsoft for copyright infringement, seeking destruction of models trained on their works. Authors and publishers are also fighting over how to split Anthropic's $1.5 billion settlement.
Agentic misuse grows. AI agents are being used to file complaints and applications at scale, with UK housing ombudsman complaints doubling and CFPB complaints up 5x. A lawyer was fined for filing a brief with ChatGPT-fabricated witnesses, and Abliteration.ai now sells API access to safety-abliterated models, lowering the barrier to misuse.
- → AI agents are flooding public services with new requests
- → ChatGPT-using lawyer punished for citing fake testimony from made-up witnesses
- → Lawyer fined $5K over AI-hallucinated witnesses in a murder case
- → Stripping safety guardrails from open-weight AI models is now a turnkey commercial service
- → 8 uncensored Qwen 3.8 27B variants, one base, 167 GPU hours - Abliterlitics
Enterprise and Open-Source AI
Local inference leaps. Community optimizations bring Qwen3.8-Flash-Next to 1.2k t/s prefill on Strix Halo, ExLlamaV3 beats llama.cpp on CPU offload, Cherenkov streams experts on Apple Silicon, and LayerStoRm runs a 186 GiB quant on 96 GB VRAM. Users are achieving dramatic speedups across RTX 3080s, Strix Halo, and MacBook Air.
- → LayerStoRm open-source expert streaming: 1M context GLM-5.3-Flash [UD-Q4_K_XL] at 24.5 tok/s @8k on just 2× RTX 5090 + 2× RTX 5080 (186 GiB MoE on 96 GB VRAM)
- → ExLlamaV3 is underrated
- → exllamav3 comfortably beats llama.cpp running CPU-offloaded Qwen-3.8-Flash-Next on my setup!
- → Qwen3.8-Flash-Next on MLX-serve, 1m context is released!
- → Qwen3.8-Flash-Next in llama.cpp vs SGLang vs FreeToken: 35s vs 258s to first token at full context. My findings on new PRs coming to engines.
- → Faster than Light in Air: 8-22 tg/s Qwen3.8-Flash-Next (Q4/Q4ish) on a 32GB M4 MacBook Air
- → Qwen3.8 Flash Next now at 1.2k t/s prefill on Strix Halo
- → 3.8-27B has ruined 3.5/3.6-35B’s for me. It’s just *absurdly* superior.
- → This draft model is OP on 16 GB cards for Qwen 3.8 27b
- → I am impressed and I owe you one, Qwen 3.8 flash next (vision)!
- → Qwen3.8 Flash Next llama.cpp config tuning
- → bartowski/Qwen3.8-27B-GGUF · Hugging Face - Updated (Per-tensor layout)
Open model audits. Open-weight models including gpt-oss-20b and glm-5.1 found real vulnerabilities in public GitHub codebases, outperforming some frontier models in security audits. Google open-sourced Mantis, an agentic vulnerability scanning framework that cuts false positives and hallucinated vulnerabilities.
Enterprise agent traction. Figma built AI agents on Panther SIEM that cut complex alert resolution time by 70% and on-call pages by 20%. But Ramp data shows AI product adoption grew only 0.4% in August, Meta stopped using AI-tool usage in performance reviews after employees gamed token usage, and forward-deployed engineer units are proliferating despite unclear goals.
- → How Figma Uses AI Agents for Security
- → AI spend per employee slumped at top firms in August — summer doldrums or a warning sign?
- → Top AI spenders cut per-employee costs by nearly 10 percent in August
- → Meta drops AI usage from engineer performance reviews after "tokenmaxxing" backfires
- → Google Cloud races to catch up in the AI deployment wars with Accenture deal
- → The Rise of the Forward Deployed Engineer — and How To Do the Job Right
Business and Compute
US-China AI theft. US agencies named DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI as engaging in industrial-scale distillation of US frontier models. Anthropic says it observed nearly 200 million exchanges linked to distillation attacks by Chinese labs.
Compute and capital. Anthropic signed compute contracts worth up to $517 billion, and Nvidia is in talks to invest up to $10 billion in Anthropic's planned IPO at a $2 trillion valuation. Cognition raised $2 billion at a $48 billion valuation, and Mistral raised €3 billion at a €21 billion valuation.
- → Anthropic reportedly signs $517 billion in compute deals after Dario Amodei warned rivals about reckless risk
- → Nvidia wants to pour up to $10 billion into Anthropic's record-breaking IPO
- → Cognition hits $48B valuation, signaling investors believe AI coding is far from a winner-take-all market
- → Mistral raises €3B as sovereign AI becomes big business
That's the week in review - see you next Sunday.