Key product launches, safety talks, and policy moves shape the AI landscape this week. Companies are rolling out new speech models, AI-agent tools, and governance frameworks.
Google unveiled Gemini 3.8 Live and Live Extended Thinking, two speech-to-speech models that generate spoken responses in real time, similar to OpenAI's GPT-Live line.
Understanding real-time speech models is essential for building conversational AI that can interact naturally with users.
Source: Simon Willison
Meta introduced a server that lets developers use AI agents such as Claude, Cursor, Codex, and ChatGPT to automate WhatsApp Business configuration, template creation, testing, and troubleshooting.
Integrating AI agents into messaging platforms demonstrates how automation can reduce manual setup effort for businesses.
Source: TechCrunch AI
OpenAI confirmed ongoing safety discussions with Anthropic and Google DeepMind, focusing on risk mitigation while political pressures push for rapid development.
Staying aware of industry safety collaborations helps engineers design systems that meet emerging compliance expectations.
Source: TechCrunch AI
Salesforce introduced Koa, a large-scale reasoning model that leverages Nvidia's Nemotron architecture to handle sales, marketing, and customer-support tasks.
New reasoning models expand the toolbox for building domain-specific AI assistants that can understand and act on complex business queries.
Source: TechCrunch AI
Microsoft released a code of conduct for its AI models, prohibiting hacking, deception, and other unsafe actions while emphasizing human-centered principles.
Adhering to model conduct guidelines is becoming a baseline requirement for deploying responsible AI services.
Source: TechCrunch AI
Study the fundamentals of speech-to-speech model architectures and how to fine-tune them for low-latency applications. Then explore building AI-agent pipelines that integrate with messaging APIs and follow safety guidelines, so you can create compliant, real-time conversational assistants.
Build my AI Engineer plan