AI Reading for Friday July 5

Microsoft claims text-to-speech model achieves parity with human readers, needs as little as 3 seconds to clone someone's voice. - Microsoft

---

OpenAI demands reporters’ notes to see how much NYT copies. NYT does have a habit of re-reporting other publications' scoops without citing them. TorrentFreak

---

By meticulously delving into the intricate web of AI slop, this comprehensive study takes a deep dive into overuse of certain phrases as significant risk factors enabling its detection. - Futurism

---

ChatGPT can be as funny as the average human. PsyPost

---

AI robot commits suicide by throwing itself down a flight of stairs after a year working for a South Korean city council - Diario AS

---

An open source contract analyzer from a big law firm. - Artificial Lawyer

---

18 humanoid robots at China AI conference, including Tesla's Optimus behind glass and powered off. We wanted FSD and humanoid robots and we got rusty Cybertrucks. - South China Morning Post

---

In an experiment, AI could detect 67% of lies, whereas humans were 50/50. Also if you gave people the lie detector tool, they tended to believe it. - MIT Technology Review

---

Suno released a mobile app to make AI music. - Music Business Worldwide

---

Mind-reading AI can tell what you are looking at from brain scan recordings. Top row is what the monkey sees, bottom is reconstruction from brain scan. How long til it can tell what you are thinking? - New Scientist

---

40-foot humanoid robot performs maintenance on Japanese railroads. - the Guardian

---

Softbank finally surpasses its 2000 high after falling 90%. - CNBC

---

Follow the latest AI headlines via SkynetAndChill.com on Bluesky