Google Open Sources Large Language Model Gemma
Google recently open sourced Gemma, a large language model with similarities to the company’s Gemini family of models. Gemma comes in two sizes - 2 billion and 7 billion parameters - and is optimized to run efficiently across devices from mobile to cloud TPUs.
This is a big move by Google to spur collaboration and progress in AI research through an open model. It provides some competition to closed models like OpenAI’s GPT-3. Gemma is available on GitHub and through Google Cloud.
source: We needed Google to enter the Open-Source AI Space, and it happened!
Reddit Selling User Content to AI Company
Reports emerged that Reddit signed a deal to sell user content to an AI company for around $60 million annually. This raised concerns around monetizing user data without explicit consent.
While platforms have rights over content shared, this highlights the lack of regulations around AI access to data. Users have limited power here unless backed by community uproar. It presents a privacy risk and ethical questions around AI training data practices.
source: Uh-Oh! Reddit Agrees to Sell User Content to an AI Company!
FTC Fines Avast $16.5 Million for Selling User Data
The FTC fined antivirus company Avast $16.5 million for selling user browsing data to advertisers despite claims of protecting privacy.
Through Avast’s browser extensions, it collected detailed browsing data and searched terms which it sold through its Jumpshot subsidiary without properly notifying users. This violated user trust and consent.
The FTC banned this data selling practice and required informing affected users. It underscores deceptive privacy practices and need for transparency in data collection.
source: FTC Slams Avast with $16.5 Million Fine for Selling Users’ Browsing Data