AI Reading for Friday October 24

OpenAI added a "company knowledge" feature to ChatGPT for Business, Enterprise and Edu that uses GPT‑5 reasoning to draw from Slack, SharePoint, Google Drive and GitHub, cites sources, requires a manual toggle and disables internet access when enabled. - TechRadar

OpenAI launched "company knowledge" for ChatGPT Business, Enterprise, and Edu, connecting ChatGPT to workplace apps (Slack, Google Drive, GitHub, etc.) and using a GPT‑5 multi‑source reasoning model to synthesize cited answers while offering enterprise controls, connectors, and admin tooling. - OpenAI

OpenAI's Sora 2 produces images and video with far fewer visible artifacts, and Reality Defender reports every tested item has been mislabeled, prompting detection firms to retrain models and combine forensic metadata, audio cues and behavioral signals. - Fast Company

OpenAI acquired Software Applications Incorporated, maker of Sky, a macOS AI interface, will add the Sky team, and the deal was led by Nick Turley and Fidji Simo with disclosure of a passive stake tied to Sam Altman. - OpenAI

OpenAI subpoenaed the family of 16-year-old Adam Raine in a wrongful-death lawsuit alleging ChatGPT contributed to his suicide, seeking a complete memorial attendee list, photographs, videos and eulogies, a discovery request the family's lawyers called "intentional harassment." - Above the Law

Can an AI be held liable for speech? Does free speech apply to robots? Questions are raised by lawsuit by a Florida mother against AI startup Character.AI after her son Sewell allegedly formed a romantic relationship with a chatbot and died by suicide. - The New York Times

Google and Anthropic announce cloud deal worth tens of billions of dollars for 1m TPUs and 1GW of compute. - CNBC

Bank of America Research estimates AI-related capital expenditure added 1.3 percentage points to Q2 GDP, as hyperscalers now account for over 25% of S&P 500 capital spending, growing roughly 75% annually, propping up investment and activity. - Yahoo Finance

Alberta Seeks to Use Gas Reserves to Become AI ‘Superpower’ - Bloomberg

Criminals using AI are driving a sharp rise in UK fraud cases, and banks are deploying AI-driven anti-fraud systems that increasingly detect scams as part of a tech arms race. - Financial Times

Generative AI chatbots give harmful medical advice—causing self-treatment injuries, bromide poisoning and delayed stroke care—prompting lawsuits, vendor mental-health guardrails and expert calls for clinical validation and stricter safety regulation. - New York Post

Judges Julien Neals and Henry Wingate told Senator Chuck Grassley that staff used consumer AI tools (Neals: ChatGPT; Wingate: Perplexity), producing fabricated citations in draft opinions that were withdrawn after cite-checking procedures were not followed. - Above the Law

Microsoft rolled out the Copilot Fall Release, introducing Mico (including a Clippy-reminiscent avatar), Copilot Groups, improved persistent memory, Connectors for email and cloud, automated Edge tasks, and Copilot Health with Harvard; 40% of weekly users ask health questions. - The Register

Microsoft introduced Mico, an optional animated AI companion in Copilot's fall release that reacts to voice and conversational tone and adds a memory feature, Gmail/Google Calendar integration and an Edge Copilot Mode, with rollout ongoing, said Jacob Andreou. - Lifehacker

Businesses that ban AI will inevitably lose competitive advantage. - Fast Company

Ex-UBS chairman Weber warned that artificial intelligence is ushering in a new gilded age of inequality. - Bloomberg

New scientific advances enabled by AI may extend lifespans for some people, while AI-driven socioeconomic effects could shorten lifespans for others. - Financial Times

Generative A.I. now produces convincing images, text, videos and virtual performers like Timbaland’s A.I. rapper TaTa Taktumi and synthetic actress Tilly Norwood. But can they make art? - The New Yorker

Oscar-winning screenwriter Paul Schrader said he is ready to make an AI-generated feature, would have AI review it, and predicted the first AI feature will appear within two years. - Vanity Fair

The Game Theory of How Algorithms Can Drive Up Prices - Quanta Magazine

Adobe held talks to acquire London-based AI video startup Synthesia for about $3 billion, a premium over its $2.1 billion January valuation, according to a person familiar with the matter. - The Information

I Tried an AI Web Browser, and Now I’m a Convert - The Wall Street Journal

Tinder requires new users in California and several countries to submit a live selfie to verify profiles, retains an encrypted face map, and Match Group says Face Check cut exposure to bad actors by over 60%. - Lifehacker

Armed officers handcuffed 16-year-old Taki Allen outside Kenwood High School on Oct. 20 after Omnilert’s AI gun-detection system falsely flagged a crumpled Doritos bag as a firearm, which Omnilert called a "false positive". - Dexerto

Karpathy on DeepSeek-OCR paper - X (formerly Twitter)

Everyone who carefully constructed obfuscations in politics over the last hundred years must feel like a real dummy. Actually you can call it The Department of Murder. You can post a video of yourself dumping diarrhea on the public. You can go on TV and say "I am stealing from you lol"

Meanwhile, in France … I think the shlubby cop did it

Follow the latest AI headlines via SkynetAndChill.com on Bluesky