#132 - FraudGPT, Apple GPT, unlimited jailbreaks, RT-2, Frontier Model Forum, PhotoGuard

Our 132nd episode with a summary and discussion of last week's big AI news!

Read out our text newsletter and comment on the podcast at https://lastweekin.ai/

Email us your questions and feedback at [email protected]

Timestamps + links:

• (00:00) Intro / Banter

• (01:36) Response to listener comments / corrections

• Tools & Apps • (05:05) OpenAI Quietly Shuts Down Its AI Detection Tool • (07:45) New AI Tool 'FraudGPT' Emerges, Tailored for Sophisticated Attacks • (10:55) JetBrains IDE update previews “deeply integrated” AI Assistant • (14:35) No More Paperwork? Amazon AI Tool Transcribes Patient Visits for Doctors • (16:11) Photoshop’s new generative AI feature lets you ‘uncrop’ images • (19:11) Wayfair’s AI tool can redraw your living room and sell you furniture •

• Applications & Business • (21:25) Apple Tests ‘Apple GPT,’ Develops Generative AI Tools to Catch OpenAI • (28:10) Facing more nimble rivals, OpenAI won’t bend … yet • (32:00) OpenAI's head of trust and safety steps down • (33:15) Google turns to AI in the race to dub YouTube • (35:28) Samsung extends cut in memory chip production, will focus on high-end AI chips instead • (36:50) Microsoft to supply AI tech to Japan government, Nikkei reports • (38:20) Protect AI raises $35M to build a suite of AI-defending tools •

• Projects & Open Source • (41:56) Why Meta is giving away its extremely powerful AI model • (49:25) Llama and ChatGPT Are Not Open-Source • (52:15) Hugging Face, GitHub and more unite to defend open source in EU AI legislation •

• Research & Advancements • (55:45) AI researchers say they've found 'virtually unlimited' ways to bypass Bard and ChatGPT's safety rules • (01:03:48) RT-2: New model translates vision and language into action • (01:09:50) (Ab)using Images and Sounds for Indirect Instruction Injection in Multi-Modal LLMs • (01:13:26) Retentive Network: A Successor to Transformer for Large Language Models • (01:18:50) STEVE-1: A Generative Model for Text-to-Behavior in Minecraft • (01:20:42) Brain2Music: Reconstructing Music from Human Brain Activity •

• Policy & Safety • (01:23:48) Major generative AI players join to create the Frontier Model Forum • (01:29:14) Cleaning Up ChatGPT Takes Heavy Toll on Human Workers • (01:32:46) America Already Has an AI Underclass • (01:35:04) AI leaders warn Senate of twin risks: moving too slow and moving too fast • (01:38:35) The Robots We Were Afraid of Are Already Here •

•

• Synthetic Media & Art • (01:42:30) This new tool could protect your pictures from AI manipulation •

• (01:44:16) Outro

Release date

Lydbog: 8. august 2023