Welcome back to Daily Zaps, your regularly-scheduled dose of AI news ⚡️ 

Here’s what we got for ya today:

• 🐍 Python remains the top AI job skill

• 🧑‍💻 OpenAI launches new models in API

• 🧠 Apple plans to train its AI on your data

• 🗣️ OpenAI considering taking on Twitter/X

Let’s get right into it!

Sign Up | Advertise | Tools

CAREERS

Python remains the top AI job skill

The 2025 AI Index Report from Stanford shows that AI jobs are growing again after a brief slowdown, with U.S. job postings requiring AI skills rising to 1.8% in 2024. Globally, demand is also increasing, particularly in Singapore, Luxembourg, and Hong Kong. While some executives foresee AI reducing their workforce, others expect growth, and studies suggest tools like GitHub Copilot may boost hiring while shifting skill demands toward broader, more human-centric abilities.

Python remains the top AI job skill, though interest in areas like generative AI has surged. However, a persistent gender gap exists, with women underrepresented in AI roles. Meanwhile, high training costs have pushed cutting-edge AI research toward industry rather than academia, prompting calls for greater academic funding and cross-sector collaboration to support future workforce needs.

STARTUPS

OpenAI launches new models in API

OpenAI has launched three new API models—GPT‑4.1, GPT‑4.1 mini, and GPT‑4.1 nano—delivering major improvements over GPT‑4o in coding, instruction following, and long-context comprehension, with support for up to 1 million tokens and a refreshed June 2024 knowledge cutoff.

GPT‑4.1 leads in benchmarks like SWE-bench Verified (54.6%), Scale’s MultiChallenge (38.3%), and Video-MME (72.0%), outperforming GPT‑4.5 and GPT‑4o while offering lower latency and cost. GPT‑4.1 mini matches or exceeds GPT‑4o in intelligence with 83% lower cost and half the latency, while GPT‑4.1 nano is optimized for speed and cost, excelling at lightweight tasks.

These models improve code generation, especially in diff formats, and enhance the ability to power autonomous agents. GPT‑4.5 Preview will be deprecated by July 14, 2025, as developers are encouraged to adopt GPT‑4.1 for better performance and efficiency.

BIG TECH

Apple plans to train its AI on your data

Apple is addressing the challenge of improving its AI capabilities without compromising user privacy by using synthetic data and a method called differential privacy. Unlike competitors like OpenAI, Google, Microsoft, and Meta—who often rely on real user data for training—Apple generates synthetic content using its own language model and refines it with anonymized insights from user devices.

With differential privacy, Apple compares synthetic data embeddings with embeddings from a small, local sample of user emails (from users who opt in), without ever accessing the original content. This approach allows Apple to enhance its AI models, such as email summarization in Apple Mail, while maintaining its commitment to user privacy.

STARTUPS

OpenAI considering taking on Twitter/X

OpenAI is exploring the idea of launching a social network to rival Elon Musk’s X and Meta’s Instagram, inspired by the viral success of its latest image-generation tool introduced in March. While the project is still in early development, the popularity of AI-generated images—especially anime-style edits—has overwhelmed OpenAI’s servers and driven significant user engagement.

CEO Sam Altman acknowledged the strain on GPU resources and announced temporary usage limits. The potential platform comes amid rising competition in generative AI, a legal feud between Altman and Musk, and OpenAI’s recent $40 billion funding round at a $300 billion valuation.