- The Business Of AI
- Posts
- 🤖 3 ways AI chatbots are a security disaster
🤖 3 ways AI chatbots are a security disaster
Plus: Meet AutoGPT
🎨
AI Image of The Week
Engine: Midjourney v5
🧠
AI and Machine Learning
Meet AutoGPT, the autonomous GPT-4 tool revolutionizing AI (2 minute read)Auto-GPT is a new development enabled by GPT-4, which has been considered a game-changer despite early concerns. It began with an open-source Python application of the same name, which has shifted attitudes towards Artificial General Intelligence (AGI) by demonstrating its potential. Auto-GPT is an autonomous experiment that aims to create a future in which AI models like GPT can independently perform tasks and achieve objectives without human intervention. As an open-source project, Auto-GPT aims to make GPT-4 entirely self-governing and has gained global popularity in recent times.
Italy will lift ChatGPT ban if OpenAI fixes privacy issues (2 minute read)Italy's data protection authority, the Guarantor for the Protection of Personal Data (GPDP), has stated that it is willing to revoke its ban on OpenAI's ChatGPT if the company meets specific requirements. The ban was put in place as part of an ongoing investigation into whether the chatbot violated Italy's data privacy laws and the EU's General Data Protection Regulation (GDPR). The GPDP was concerned that ChatGPT could recall and emit personal information, including phone numbers and addresses, from input queries and expose minors to harmful content. To lift the ban, OpenAI must enforce rules protecting minors and personal data by 30th April 2023, including notifying users of data processing practices on its website and requiring users to confirm they are 18 years old or older before using the software.
Three ways AI chatbots are a security disaster (6 minute read)AI language models are currently the most exciting development in tech, but they may also be the source of a significant problem: they are incredibly easy to misuse as phishing or scamming tools, and there is no known solution. Tech companies are rushing to integrate these models into various products, from travel booking to note-taking. However, the way these products work creates new risks, including the potential to leak private information and assist criminals with phishing, spamming, and scamming. Experts predict a looming security and privacy "disaster."
💼
Business
The AI race is just getting started (2 minute read)OpenAI's early success in the field of generative AI does not guarantee long-term dominance, as "first movers" in past technology leaps have stumbled and been overtaken by larger rivals. The fate of Netscape, which pioneered web browsers in the 1990s before being crushed by Microsoft, serves as a cautionary tale. OpenAI's future is uncertain and faces significant obstacles, including its reliance on vast cloud computing resources controlled by Amazon, Microsoft, and Google.
Amazon AWS expands generative AI efforts with Bedrock and CodeWhisperer updates (4 minute read)Amazon Web Services (AWS) is expanding its generative AI services, including Amazon Bedrock, a set of foundation model AI services. The launch includes models developed by AWS and those from partners AI21, Anthropic, and Stability AI. AWS is also releasing the general availability of Amazon EC2 Inf2 cloud instances powered by AWS Inferentia2 chips, and making the Amazon CodeWhisperer generative AI service for code development free for individual developers. Bratin Saha, VP of ML and AI Services at Amazon, stated that the goal of these launches is to democratize the use of generative AI. AWS is expanding its AI services as cloud rivals Microsoft Azure and Google Cloud also push their own services.
Microsoft integrates Bing into its keyboard SwiftKey app on Android and iOS (2 minute read)Microsoft has announced that its Bing chatbot has been integrated into its SwiftKey smartphone keyboard app for both Android and iOS. The integration enables users to chat with the bot directly from their mobile keyboard and search for things without switching apps. Bing's integration with SwiftKey offers three major features, including Chat, Tone, and Search. With the Chat feature, users can access Bing for more detailed queries, such as finding a clever pun to respond to someone's message or searching for the ideal dessert that pairs well with a specific dish when planning a party with friends on mobile.
❓
Misc
AI clones teen girl’s voice in $1M kidnapping scam: ‘I’ve got your daughter’ (4 minute read)An Arizona mother claims that scammers used AI to clone her daughter's voice and demand a $1 million ransom from her as part of a terrifying new voice scheme. The use of AI in phone scams is becoming increasingly common, with scammers using "caller-ID spoofing" schemes to claim that they've taken the recipient's relative hostage and will harm them if they aren't paid a specified amount of money. The woman received a call from an unfamiliar phone number, which she almost ignored before the AI-generated voice convinced her otherwise.
Measuring trends in Artificial Intelligence (10 minute+ read)The production of significant machine learning models has shifted from academia to industry since 2014, with 32 industry-produced models in 2022 compared to three from academia. Industry has an advantage due to greater access to resources such as data, compute, and money. Meanwhile, AI models are increasingly being used to accelerate scientific progress, as seen in their contributions to hydrogen fusion, matrix manipulation, and antibody generation.
🐥
Best of Twitter
Big batch of announcements today from Amazon, Stability AI, Google and Meta.
Here's what happened and how to make sense of them:
— Pete (@nonmayorpete)
9:00 PM • Apr 13, 2023
Thanks for reading, if you enjoyed, tell your friends!