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

Here’s what we got for ya today:

• 🤔 ChatGPT will remember everything you tell it

• 🧑‍💻 Firebase Studio - Google takes on Cursor

• 🔐 Sec-Gemini v1, AI model for cybersecurity

• 🤝 Google’s Agent2Agent protocol

Let’s get right into it!

Sign Up | Advertise | Tools

STARTUPS

ChatGPT will remember everything you tell it

OpenAI has begun rolling out a new memory feature in ChatGPT that enables the chatbot to personalize responses based on users' past conversations, enhancing the relevance and fluidity of interactions across text, voice, and image-generation. Initially available to ChatGPT Pro and Plus subscribers—excluding users in the U.K., EU, and a few other regions due to regulatory reviews—the feature will be enabled by default for those who previously used memory.

Users can view, manage, or disable saved memories in settings, or opt for Temporary Chats that don’t store information. While similar to a recent feature introduced by Google’s Gemini, OpenAI’s update aims to make ChatGPT feel more personal and intuitive without requiring repeated input, though privacy-conscious users may have concerns.

BIG TECH

Firebase Studio - Google takes on Cursor

Google has quietly launched Firebase Studio, a cloud-based, AI-powered integrated development environment (IDE) that allows users to build full-stack apps using natural language prompts and images, without needing extensive coding knowledge.

Positioned as Google’s answer to the growing “vibe coding” trend popularized by tools like Cursor AI, Firebase Studio supports a wide range of languages and frameworks including React, Next.js, Angular, Flutter, and more. Users can quickly prototype apps by simply describing their ideas, and then refine the code directly within the IDE. While still requiring a Gemini API key, Firebase Studio represents Google’s latest move to compete in the AI-driven app development space.

BIG TECH

Sec-Gemini v1, AI model for cybersecurity

Google has unveiled Sec-Gemini v1, an experimental AI model aimed at strengthening cybersecurity defenses by helping analysts detect threats, analyze incidents, and understand vulnerabilities more quickly and accurately. Developed by Google’s cybersecurity experts Elie Burzstein and Marianna Tishchenko, Sec-Gemini v1 leverages real-time data from trusted sources like Google Threat Intelligence, Mandiant, and the Open Source Vulnerabilities database to provide deeper insights into security threats and attacker tactics.

Unlike other AI tools, it contextualizes vulnerabilities and identifies root causes, achieving benchmark-beating results—outperforming GPT-4 and Claude on key cybersecurity tests. While still in research phase, Sec-Gemini v1 could shift the balance in the ongoing battle between defenders and attackers, as Google positions it as a force multiplier in an increasingly complex digital threat landscape.

Interested in testing Sec-Gemini v1? Google is taking requests via this form.

BIG TECH

Google’s Agent2Agent protocol

Google has introduced Agent2Agent (A2A), a new open protocol designed to enable AI agents to collaborate seamlessly across different platforms, frameworks, and vendors, with support from over 50 major technology and consulting partners. As AI agents become central to automating complex enterprise workflows—from IT support to supply chain planning—interoperability becomes critical.

A2A allows agents to securely exchange information, coordinate actions, and manage tasks across siloed systems, enhancing productivity while reducing long-term costs. Built on familiar standards like HTTP and JSON-RPC, A2A is secure by default, supports long-running and multi-modal tasks, and provides a standardized method for agent communication and capability discovery. Real-world applications, such as collaborative candidate sourcing, highlight its transformative potential. Google is releasing A2A as open source with plans for a production-ready version later this year, aiming to drive a future of flexible, interoperable, and intelligent agent ecosystems.