TechCrunch Summary (2023-04-28 12PM)

Source: Kenyan court paves way for lawsuit alleging Facebook played role in fuelling Ethiopia’s Tigray conflict

Summary
Ethiopians suing Meta have been allowed to serve the social media site outside Kenya for failing to have physical offices locally and to stop amplifying violent posts during the Tigray War. The legal petition blames Facebook’s algorithm for amplifying hateful and violent posts that could have endangered lives. The petitioners seek to compel Meta to stop viral hate on Facebook, ramp-up content review in Kenya and create a $1.6 billion compensation fund.

Fact

💻 Meta has no physical offices in Kenya, which is hampering the legal suit.

🤝 Facebook’s moderation hub in Kenya had only 25 moderators responsible for Amharic, Tigrinya, and Oromo content, leaving out 82 other languages with no personnel to moderate.

⚖️ The petitioners seek to create a $1.6 billion compensation fund for victims of Facebook’s content moderation failure.

📉 Facebook’s algorithm allegedly amplified hateful and inciteful posts, causing more interactions and damaging lives.

🚮 Facebook is accused of having “under-invested” in human content review at the moderation hub in Kenya, risking lives.

---

Source: ChatGPT resumes service in Italy after adding privacy disclosures and controls

Summary
OpenAI has reintroduced its chatbot, ChatGPT, in Italy after responding to concerns raised by the country’s data protection authority (DPA) that the service breached the EU’s GDPR. OpenAI now requires consent from a parent or guardian to use the service for individuals aged under 18 or 13 respectively. The firm has also expanded its privacy policy and provided information about the personal data used to train its algorithms. European regulators have launched a task force to assess the service.

Fact

🤖 OpenAI has reintroduced its ChatGPT chatbot to users in Italy after addressing compliance concerns raised by the country’s data protection authority

🇮🇹 Users on Italian IP addresses must confirm they are over 13 or 18 by clicking a button before accessing the service (with or without a parent or guardian’s consent)

👨‍⚖️ The data protection authority required OpenAI to make modifications to the legal basis of the service, add age-gating measures and provide more information about data privacy and opt out rights, among other steps

---

Source: #NotMyAI and other TC news

Summary
Snapchat launched My AI, a generative chatbot, to its 750 million monthly users. TechCrunch reporter Amanda Silberling discusses making friends with robots online.

Facts

🤖 My AI is a generative chatbot from Snapchat.

👥 The chatbot is available to Snapchat’s 750 million monthly users.

💬 The TechCrunch Podcast episode explores making friends with robots online.

---

Source: Amazon is developing an improved LLM to power Alexa

Summary
Amazon is developing a new large language model(LMM) to enhance Alexa’s capabilities as a personal assistant. The CEO, Andy Jassy believes that the move towards more generalized and capable LMMs will make underlying models more effective and help achieve Amazon’s goal of building the “world’s best personal assistant.”

Facts

🤖 Amazon is building a new, improved large language model to enhance Alexa’s capabilities for use as a personal assistant.

📈 Such improvements are “difficult…across a lot of domains,” but Jassy believes the new LLM will “accelerate the possibility of building that world’s best personal assistant.”

🤝 Amazon has invested in AI and LMMs for years and predicts LMMs will have significant business models.

👨‍💼 Amazon recently launched Bedrock, a way to build generative AI-powered apps via pre-trained models from notable startups including AI21 Labs, Anthropic, and Stability AI.

👩‍💼 Apple is also developing LMM-based improvements for Siri, with Google likely doing something similar for Assistant.

---