- The Shift
- Posts
- Apple Tests AI Giants
Apple Tests AI Giants
Plus, 🎬 Meta strikes licensing deal with Midjourney for AI models, Control AI Videos Frame-by-Frame with Kling 2.1, and more!
Hello Readersđź‘€
Curious about the biggest AI moves today? You’re in the right place. In today’s edition, we have:
🍏 Apple Eyes Anthropic, OpenAI, and Google to Reinvent Siri
🎬 Control AI Videos Frame-by-Frame with Kling 2.1
🎨 Meta Partners with Midjourney on AI Image & Video Models
🔨Tools you cannot miss
🍏 Apple Eyes Anthropic, OpenAI, and Google to Reinvent Siri
Apple is weighing whether to power Siri with third-party AI, testing Anthropic’s Claude, OpenAI’s ChatGPT, and Google’s Gemini on its private servers. Bloomberg reports Apple is still weeks away from choosing whether to stick in-house or go external.
The Shift:
1. Custom Gemini Already in Testing - Apple had Google build a tailored Gemini model designed to run on its own private cloud servers. Training is already underway, allowing Apple to test Gemini’s performance while maintaining its strict privacy framework.
2. Two Competing Siri Versions - Apple is developing “Linwood” with its own Foundation Models and “Glenwood” powered by external AI like Claude, ChatGPT, or Gemini. Both versions are being tested in parallel to evaluate performance.
3. Anthropic & OpenAI Still in Play - Beyond Google, Apple has discussed partnerships with Anthropic and OpenAI. ChatGPT is already embedded into Siri for some query handling, but Apple is considering deeper integration if third-party tech wins out. Executives say they’re several weeks away from a decision on both the partner and the long-term direction of Siri.
Apple is openly acknowledging that its own AI may not be enough to revive Siri. Choosing a partner like Anthropic, OpenAI, or Google could accelerate Siri’s evolution by years, finally putting it on par with rivals.
🎬 Control AI Videos Frame-by-Frame with Kling 2.1
Marketers can now generate hyper-precise AI videos using just two key visuals: the starting and ending frames. Kling 2.1’s “Image-to-Video” tool fills in everything between, letting you guide the narrative while AI handles the animation.
How to Use Kling 2.1’s Start/End Frame Control
Go to Kling 2.1 → Select AI Video Generator
Choose Image to Video
Upload your start image (first frame of the scene)
Upload your end image (where the clip should finish)
(Optional) Add a prompt for tone, style, or action
Click Generate
Kling handles the in-between, you just guide the start and finish.
🎨 Meta Partners with Midjourney on AI Image & Video Models
Meta announced a licensing deal with Midjourney to integrate its image and video generation tech into future AI products. The move signals Meta’s push to compete with OpenAI’s Sora, Google’s Veo, and other leaders in generative visuals.
The Shift:
1. Meta Gains Midjourney’s Tech Edge - The deal gives Meta access to Midjourney’s signature image and video models, known for realism and unique style. Meta already runs Imagine and Movie Gen, but Midjourney’s tools could accelerate product quality.
2. Bigger AI Race Moves - The Midjourney deal follows Meta’s $14B investment in Scale AI, its acquisition of voice startup Play AI, and a $100M hiring spree for AI researchers. Zuckerberg reportedly also held talks with Elon Musk about joining his $97B bid for OpenAI, though Meta declined.
3. Copyright & Independence Questions - Midjourney, valued at ~$200M in annual revenue, has never taken outside funding and runs on a subscription model. Holz stressed that the company will stay independent even while collaborating with Meta. Still, lawsuits from Disney and Universal over training data put both Midjourney and Meta under growing copyright scrutiny.
This deal underscores how the AI race is shifting toward high-quality generative visuals as a competitive frontier. For Meta, partnering with an independent player like Midjourney helps close the gap with rivals while diversifying its AI strategy. It also highlights how copyright battles are unlikely to slow Big Tech’s push into next-gen creative tools.
🔨AI Tools for the Shift
✍️ WriteHybrid – Paste AI text and instantly get humanized content. Perfect for creators, marketers, and businesses looking for natural-sounding writing.
🎨 Nano Banana AI – Edit, adjust, and optimize your images quickly with AI. Ideal for designers, marketers, and social media content creators.
⚡ Stakly – Build production-ready web apps in hours with AI. A developer-friendly platform for rapid, reliable app creation.
📊 Mobser – AI-powered data analytics for smart reports and visualizations. Make better decisions with automated insights.
🎙️ WriteVoice – Stop typing, start talking. Convert your voice into text seamlessly across all apps.
🚀Quick Shifts
🤖 Elon Musk’s xAI has trademarked "Macrohard," a tongue-in-cheek rival to Microsoft, as a purely AI-driven software company. The venture includes multi-agent systems via Grok and is reportedly real.
🚀 xAI has released the Grok 2.5 model weights on Hugging Face, its top-performing model from last year, and plans to open-source Grok 3 in six months. The custom license includes anti-competitive clauses.
⚠️ OpenAI warned investors against unauthorized SPV deals for its equity, saying such sales won’t be recognized. Anthropic is taking a similar stance, forcing VCs to invest directly instead of via SPVs.
🎥 Baidu has launched MuseStreamer 2.0, producing 1080p, 10-second videos across Turbo, Pro, and Lite tiers. With native Chinese audio and an 89.4% VBench score, access begins via HuiXiang in China.
🧬 OpenAI and Retro Biosciences built a custom AI to redesign stem cell proteins, boosting efficiency 50x and reversing aging markers, a landmark for domain-trained AI in biology, chemistry, and materials science.
Reply