AI Reading for Tuesday August 27

Non-paywalled discussion of Strawberry/Orion - THE DECODER

---

OpenAI said to have demoed its next-gen 'Strawberry' model to the Feds, to use it to power an 'Orion' model for release this fall. - The Information

---

Robocrop field worker robot only picks the ripe strawberries. - the Guardian

---

OpenAI looks for someone to investigate its employees. Not sure if I should worry OpenAI is turning itself into a surveillance state against whistleblowers, or that insiders are stealing my ill-advised queries with proprietary information for nefarious purposes. - Business Insider

---

OpenAI, other Big Tech support CA watermark bill - TechCrunch

---

Elon Musk comes out in favor of the controversial CA AI safety bill, absence of Grok guardrails notwithstanding - X (formerly Twitter)

---

Musk and Cuban engage in spat as X will surface more 'banger' small accounts i.e. edgelords Musk approves of - Business Insider

---

A deepfake Elon livestream - Reddit

---

There's an Nvidia earnings watch party at an NYC bar tomorrow. File under 'signs of a top' - Business Insider

---

Short-seller Hindenburg takes aim at SuperMicro. - Yahoo Finance

---

Buy now pay later firm Klarna aims to halve headcount using AI. - FT

---

Scale.ai, which like Amazon's Mechanical Turk uses human workers to annotate AI training data sets, and has an improbably high valuation, lays off over 1,000 workers. - Inc.

---

Now the CIA really can implant a chip to read your mind. - Interesting Engineering

---

Legislation may be needed to prevent the use of AI to engineer a pandemic. - Science

---

Cerebras said it launched the world's fastest AI inference service at 1800 tokens per second. - SiliconANGLE

---

Cerebras, Groq, and D-Matrix aim to dethrone Nvidia. - FT

---

Larry Page said what now? - X (formerly Twitter)

---

A new Beethoven 10th symphony fleshed out by AI from scribbles he left behind, and played by a real symphony orchestra. - Washington Post

---

Antrhopic published its system prompts. - TechCrunch

---

AI doomers had a moment, did they waste their shot? - The Atlantic

---

Updated $10/month Alexa service coming in October. - Washington Post

---

Gannett shutters site that was publishing AI-generated product reviews under fake personas. - The Verge

---

Everybody gonna have their own AI personal assistants mailing each other now - Reddit

---

Champions League says they will use 'AI' for a draw in a new league structure with 8 games against 8 different teams, 4 home away, instead of group stage with teams playing each other both home and away. Could be more interesting and fair, without 'groups of death' if they don't mess it up. - Goal

---

2 can play that game: China restricts exports of critical raw materials, spiking prices. - FT

---

Lawyers will start asking cops in court, did you write this arrest report or did an AI write it? - pennlive

---

Model collapse: when AI output is a threat to AI itself.

---

Will AI destroy the planet or save the planet? - NY Times

---

Dear ChatGPT, I hate you - Reddit

---

Follow the latest AI headlines via SkynetAndChill.com on Bluesky