OpenAI Releases Codex

Grok gets smarter with internal prompt updates, scientists use AI to hide spy messages, and Coinbase investigates a new customer data breach.

Welcome back to your daily dossier talking AI, robotics, crypto and the future of tech.

In today’s edition we are tackling the following:

🧠 OpenAI launches Codex, a coding agent embedded in ChatGPT for software tasks. 🤖 xAI revamps Grok’s prompts to improve reasoning, style, and conversation context. 🕵️‍♂️ Scientists create AI that hides messages in images to bypass cybersecurity.
🌾 Centipede-style bot targets farm weeds while crawling through fields with precision. 💸 Coinbase confirms hackers accessed sensitive customer data through phishing.

Flex your product muscles with Product for Engineers.

Product for Engineers is PostHog’s newsletter dedicated to helping engineers and founders improve their product skills. With over 50k subscribers, it is a trusted source bridging the gap between product and engineering, helping you ship products users love.

Issues cover curated advice on building great products, lessons (and mistakes) they’ve learned from building PostHog, and deep dives on the culture and strategies of top startups.

*This is sponsored content

AI

  • OpenAI launches Codex, an AI coding agent, in ChatGPT

    • Codex enables ChatGPT to autonomously write, edit, and debug code, taking on full software engineering tasks end-to-end.

    • The agent uses a live file system, task queue, and review interface to mirror real-world developer workflows.

    • Codex is currently available in alpha to pro users, with broader rollout planned as capabilities improve.

  • xAI's Grok system now runs on optimized internal prompts

    • xAI redesigned Grok’s prompt system to improve speed, context retention, and performance across chat experiences.

    • The update includes multi-stage prompting chains, enhanced reasoning, and personalized behavioral tuning.

    • xAI says the changes allow Grok to better emulate human logic while maintaining efficiency and style.

  • Scientists use AI to encrypt secret messages invisible to cybersecurity systems

    • Researchers developed an AI method that hides data in images in ways undetectable by conventional security tools.

    • The system uses adversarial steganography, training AI to embed messages that evade machine and human detection.

    • This raises new possibilities—and risks—for espionage, secure communication, and AI-enabled cyber threats.

ROBOTICS

CRYPTO

  • Coinbase says cyber criminals stole account data from some customers

    • Coinbase disclosed that hackers accessed sensitive account data from a subset of users in a recent breach.

    • Attackers reportedly used social engineering to trick employees and gain unauthorized internal access.

    • The company is investigating and working with law enforcement while strengthening internal security protocols.

  • World Liberty’s stablecoin now available on multiple networks through Chainlink

    • Chainlink’s interoperability protocol now supports World Liberty’s stablecoin across Ethereum, Solana, and other blockchains.

    • This integration enhances cross-chain liquidity and expands the stablecoin’s usability in DeFi and payments.

    • Developers say it marks a step forward for seamless digital asset movement between ecosystems.

  • Nasdaq-listed healthcare firm plans $1 billion bitcoin acquisition

    • A public healthcare company revealed plans to acquire $1B in bitcoin, citing long-term value preservation.

    • The firm sees crypto as a hedge against fiat volatility and rising global inflation.

    • Analysts are watching closely as this signals growing institutional adoption beyond tech or finance.

Your startup is scaling fast. Why isn’t your SOC 2?

Delve allows founders and compliance teams to auto-capture all screenshots for SOC 2, HIPAA, ISO, and more to get compliant in 15 hours or less.

Don't let outdated compliance methods cripple your startup's liftoff. If your current provider is too slow, we’ll even migrate you off. Book a 30-minute demo and blast off with $1,000 off using code MEMORANDUM1K.

(PS: Get compliant with Delve and we’ll celebrate with AirPods Pro on us.)

*This is sponsored content

CONTENT

  • Startup Ideas You Can Now Build With AI

    • Y Combinator partners explore how AI makes previously impossible startup ideas—like full-stack law firms—suddenly viable.

    • Personalized education, recruiting, and vertical SaaS tools now benefit from scaled capabilities enabled by LLMs.

    • They emphasize second chances: many failed ideas now work due to better distribution, infra, and smarter AI.

  • OpenAI Launches Codex: An Autonomous Programming Agent

    • Codex autonomously builds features, fixes bugs, and manages pull requests, acting like a junior engineer on command.

    • OpenAI envisions it as part of a broader super-assistant that integrates seamlessly into user workflows.

    • Codex uses parallel task execution and a hands-off design, encouraging an ‘abundance mindset’ for managing agents.

  • Who’s Coding Now? AI and the Future of Software Development

    • a16z discusses how AI tools reshape programming, pushing coding into a high-level, prompt-based abstraction layer.

    • Developers still need deep understanding of logic and architecture to manage non-deterministic agent behavior.

    • Formal languages won’t disappear, but how and when we use them will shift as vibe coding becomes the norm.

AND MORE…

  • AI video startup Moonvalley lands $53M, according to filing

    • Moonvalley, a stealthy AI video company, quietly raised $53 million in a previously unannounced funding round.

    • The startup is working on generative video tools and has ties to former Google and YouTube executives.

    • Investors include top venture firms betting on next-gen content creation powered by AI.

  • Microsoft CEO Satya Nadella on his AI efforts and OpenAI partnership

    • Nadella emphasized Microsoft’s long-term vision of AI embedded across all products and workflows.

    • He reaffirmed support for OpenAI, viewing the partnership as a foundational pillar of Microsoft’s AI dominance.

    • Nadella acknowledged growing scrutiny, stressing the need for AI regulation and responsible development.

  • Pope Leo XIV voices concerns over artificial intelligence

    • Pope Leo XIV warned that AI could deepen inequality and erode human dignity without careful moral oversight.

    • He called for global cooperation on ethical standards to guide the development of intelligent systems.

    • His message comes as the Vatican joins ongoing discussions around tech and human-centered design.

P.S. Want to collaborate?

Here are some ways.

  1. Share today’s news with someone would would dig it. It really helps us to grow.

  2. Let’s partner up. Looking for some ad inventory? Cool, we’ve got some.

  3. Deeper integrations. If it’s some longer form storytelling you are after, reply to this email and we can get the ball rolling.

How was today's email?

Login or Subscribe to participate in polls.