The Agentic Newscast provides an overview of recent developments in the AI landscape, highlighting significant moves by major tech companies like Apple, which is restructuring its AI teams, and Foxconn, which is building new facilities in Houston for Apple and Nvidia hardware. The text also discusses the hardware race, specifically Huawei's new chip aiming to compete with Nvidia's H100. Additionally, the sources cover updates to AI models, including OpenAI's patch for GPT-4o and the retirement of GPT-4. Finally, they touch on potential legal risks associated with AI agent systems and spotlight current AI agent frameworks popular with developers.Apple starts dismantling its central AI org – robotics R&D moves to the Hardware group and Siri gets its own VP under Craig Federighi, a clear push toward deeper on-device “personal agents.” (What Is Happening With Apple’s AI Team? Siri and Robotics Shifts Mark a Breakup - Bloomberg)

Microsoft sets 27 Apr preview date for “Security Copilot Agents.” The autonomous add-ons will triage phishing alerts, enforce Intune policies and more, offloading grunt work from SOC teams. (Microsoft Gives Security Copilot Some Autonomy)

Reuters deep-dive flags new liability landmines for AI agents, from mis-aligned decisions to contracts signed without human sign-off; urges firms to treat agents as “high-risk actors” requiring formal governance. (AI agents: greater capabilities and enhanced risks - Reuters)

MIDI Agent brings AI composition straight into any DAW – the VST/AU plugin generates, transcribes and extends MIDI with ChatGPT, Claude, Gemini and more, connecting AI to digital audio workstations like Ableton, Logic, FL Studio. (MIDI Agent - KVR Audio)