Last Week in AI #132 - FraudGPT, Apple GPT, unlimited jailbreaks, RT-2, Frontier Model Forum, PhotoGuard
26 snips
Aug 8, 2023 Recent discussions highlight the emergence of FraudGPT, a tool designed for sophisticated attacks, raising concerns about AI misuse. Apple's push into generative AI is making waves as it tests Apple GPT to compete with leaders like OpenAI. Innovations in healthcare with AWS HealthScribe are revolutionizing doctor visits, while Wayfair’s AI tool lets users design their living rooms seamlessly. Additionally, the creation of PhotoGuard aims to protect images from AI manipulation, underscoring the race for ethical AI advancements.
AI Snips
Chapters
Transcript
Episode notes
AI Ransom Calls
- A listener alerted Andrei Krenkov and Jeremy Harris to Belva.ai, a service that makes AI ransom calls.
- This highlighted the ease of malicious AI use.
Podcast Navigation Tip
- Use timestamps in podcast descriptions to allow listeners to quickly find specific sections.
- This lets listeners skip parts and make listening more efficient.
AI Detection Limits
- OpenAI's AI detection tool was unreliable and quietly shut down.
- AI text detection is a losing battle against generation, highlighting the limitations of current technology.