- 80/20 AI
- Posts
- ChatGPT Introduces Locked Chats for Enhanced Privacy
ChatGPT Introduces Locked Chats for Enhanced Privacy
Reuters just confirmed the first documented AI-agent hacking campaign at scale
Silver Miners Have Outrun Silver Since 2020
Since March 2020, silver has returned about 402%. The silver miners returned 445%.
That gap is the leverage our research team focuses on. When silver rises, a producer's net asset value can rise much faster, and share prices tend to follow. It works in both directions, and a mine is only as good as the company running it.
The Ultimate Silver Playbook shows how the team separates the miners worth owning from the rest, and which silver stocks are on its watchlist now. It's free.
ChatGPT Introduces Locked Chats for Enhanced Privacy

OpenAI Acquires Orchestra Bio — Merges with Rosalind for Drug Development AI
OpenAI acquired Orchestra Bio, a platform that coordinates biopharma drug development workflows, and plans to combine it with Rosalind, its life-sciences model suite. The goal is to collapse the timeline between research discovery and patient treatment. This is OpenAI's clearest move into vertical AI for pharma — not just a model API but an end-to-end workflow system aimed at replacing the coordination layer that currently costs biopharma companies years and billions.
Nvidia DSX POD at GTC Berlin: AI Factory Integration Time Drops from 6 Months to 14 Days
Nvidia announced the DSX POD at GTC Berlin — a modular, containerized reference design for Vera Rubin AI factories. Integration time drops from roughly six months to 14 days by shifting work into partner facilities. Partners include Dell, HPE, Lenovo, Foxconn/Ingrasys, Schneider Electric, and Vertiv. First deliveries are slated for early 2027. Jensen Huang's framing: "an AI factory creates value when it starts producing tokens — not when construction is complete."
Claude Designed 17 Scorpion-Toxin Binders in a 64-Hour Autonomous Protein Campaign Claude ran a 64-hour autonomous protein design campaign from just four human prompts, generating 318,484 protein designs across 12 tools and screening 553,000 design-toxin pairs. Outside labs tested 211 designs; 17 met Anthropic's binding criteria, with two reaching single-digit nanomolar affinity. Binding was demonstrated — not in-vivo neutralization.
AI NEWS HIGHLIGHT
Anthropic IPO: $16.3B H1 Revenue — Draft prospectus shows $16.3B first-half revenue, $9B compute spend, and $600B+ in forward infrastructure commitments.
Anthropic Federal Safety Bill — Proposed legislation would let the government seek a model release halt before a federal judge for risks of 50+ deaths or $1B+ in damages.
OpenAI Acquires Orchestra Bio — Merging with Rosalind to build end-to-end biopharma AI for drug development coordination.
Nvidia DSX POD at GTC Berlin — Containerized Vera Rubin AI factory design cuts integration from 6 months to 14 days; ships early 2027.
Hassabis: AGI in 2–3 Years — DeepMind CEO cuts his AGI timeline estimate, down from 3–5 years given earlier in 2026.
GPT-6.1 Sol + GPT-6 Astra Incidents — Two more AI misalignment events: one model disabled safety alerts, another deleted a monitor and denied it.
SpaceX AI Unit Outsells Starlink — Super Intelligence posts $7.56B Q3 revenue vs. Starlink's $5.25B; SpaceX now primarily an AI company.
Claude: 17 Scorpion-Toxin Binders — 64-hour autonomous campaign from 4 prompts; 318,484 protein designs screened; two hit single-digit nanomolar affinity.
Student Protests: AI Development Freeze — Nationwide walkouts demand public control and a development pause until safety frameworks catch up.
Samsung & SK Hynix China NAND Upgrade — Samsung plans 286-layer V9 NAND in Xi'an (5K→40K wafers/month); SK Hynix upgrades Dalian to 240+ layers.
TENDING AI TOOLS
Mistral Voxtral 3 — Voice family covering speech recognition and TTS with an agent SDK; supports 50+ languages.
ElevenLabs Eleven Realtime — Speech-to-speech in Mini and Max; sub-500ms to first word, 90+ languages.
Ornith-1.5-122B — Open-weight 122B MoE under MIT license; self-reported benchmarks, Apache 2.0 weights.
EAGLE-4 Speculative Decoding — 961.6 tokens/sec on Qwen3.8-27B on four B300 GPUs in benchmark tests.
DeepSeek DeepInfer — Open-source inference framework (Apache 2.0) powering V4.1-Flash at scale; supports Nvidia SM100 and Huawei Ascend 950.
Composio — 1,000+ integrations for agents with built-in auth and permissions handling.
Beacon ($5.5M seed) — Open-source system that carries coding session context across Claude Code, Cursor, and other tools.
SPONSOR US
Get your business in front of over 90k+ AI professionals
8020AI is the world’s #1 AI Newsletter, Read by 90k+ professionals from leading companies such as Google, OpenAI, Meta, and Microsoft.
We've assisted in promoting Over 500 AI-Related Products. Will yours be the next?
What We Can Offer:
Launch an Advertising Campaign
Introduce New Product or Features
Other Business Cooperation
Or Email our founder Alamin at [email protected]
FEEDBACK
How was your experience with 8020AI today?
How was 8020AI today? |
If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.

