Google's AI-Powered Circle to Search

Welcome to this week’s AI & Tech Insider!

From smarter search experiences to groundbreaking ethical AI benchmarks and urgent data security alerts, this edition covers the latest advancements shaping our digital world.

Google’s Circle to Search just got a powerful upgrade with AI Mode, making it easier than ever to explore and get instant AI insights without leaving your app. Meanwhile, new research introduces the Flourishing AI Benchmark, a holistic approach to evaluating how well AI supports human well-being across multiple life dimensions — a critical step toward ethical AI development.

However, the risks of AI aren’t just theoretical: a serious data breach in McDonald’s AI-powered hiring bot has exposed millions of applicants’ personal information, spotlighting the importance of robust security in AI systems.

Dive in to stay informed, empowered, and ahead of the curve!

Google Enhances Circle to Search with AI Mode

Google is enhancing Circle to Search by integrating AI Mode, allowing users to ask follow-up questions and explore topics more deeply without leaving their current app. Now available on over 300 million Android devices, Circle to Search lets you long press the home button or navigation bar, then circle, tap, or gesture on your screen to get instant AI-powered insights. When an AI Overview is most useful, you’ll see it in your results, and by tapping “dive deeper with AI Mode,” you can continue your exploration with advanced reasoning and web-wide content. AI Mode is also accessible through Lens in the Google app on both Android and iOS, currently available in the U.S. and India. Additionally, Circle to Search now offers real-time gaming help—while playing a mobile game, you can identify characters or find strategies by simply circling or tapping on the screen, receiving AI Overviews and suggested videos without interrupting gameplay. Recent upgrades to AI Overviews, powered by Gemini models, make responses clearer and more visual, providing concise breakdowns and richer context for all types of searches.

Measuring AI Alignment with Human Flourishing

The paper "Measuring AI Alignment with Human Flourishing" introduces the Flourishing AI Benchmark (FAI Benchmark), a new evaluation framework designed to assess how well AI systems align with holistic human well-being across seven key dimensions: Character and Virtue, Close Social Relationships, Happiness and Life Satisfaction, Meaning and Purpose, Mental and Physical Health, Financial and Material Stability, and Faith and Spirituality. Unlike traditional AI benchmarks that focus mainly on technical performance or harm prevention, the FAI Benchmark uses 1,229 objective and subjective questions to evaluate large language models (LLMs) on their ability to support human flourishing. The methodology incorporates specialized judge LLMs and cross-dimensional scoring using the geometric mean to ensure balanced assessment. Initial results from testing 28 leading language models show that while some models perform well overall (with top scores reaching 72/100), none fully align with all aspects of human flourishing, especially in areas like Faith and Spirituality, Character and Virtue, and Meaning and Purpose. This research highlights the need for AI systems that proactively foster human flourishing, setting a new standard for ethical AI development and evaluation.

McDonald’s AI Bot Exposed Millions in Data Breach

A major security flaw in McDonald’s AI-powered hiring platform, McHire—run by Paradox.ai and featuring the Olivia chatbot—left the personal data of tens of millions of job applicants exposed to hackers. Security researchers Ian Carroll and Sam Curry discovered that basic vulnerabilities, including an administrator account with the password "123456," allowed them to access up to 64 million records containing applicants’ names, emails, and phone numbers. While Paradox.ai confirmed the breach and claimed only a fraction of records included personal data, both Paradox.ai and McDonald’s acknowledged the severity of the issue, with Paradox.ai pledging to implement a bug bounty program and McDonald’s promising stricter oversight of third-party providers. The vulnerability was reportedly fixed the same day it was reported, but the incident highlights the risks of using AI-driven systems with poor security practices in sensitive processes like hiring.

Hand Picked Video

In this video, we'll look at creating a Chrome Extension from scratch using Vibe Code.

Top AI Products from this week

• Intervo - Intervo is the open-source platform for creating sophisticated phone call, voice and chat AI agents. Go beyond the limitations of closed platforms like ChatBase and Retell.AI to instantly assist customers qualify leads, handle support and do a whole lot more!

• Permut - Permut is the complete platform to build AI agents that engage customers, collect feedback, and drive growth — across WhatsApp, SMS, Web and more. Boost retention, prevent poor experiences, and unlock real-time insights at scale.

• opencode - opencode is an open-source AI coding agent built for the terminal. It features a native TUI, multi-session support, and is model-agnostic, letting you use Claude, OpenAI, Google, or even local models.

• Sublime - Sublime is a journal that turns your thoughts into action. Once daily, reflect with a thoughtful prompt. Our AI keeps tabs on what's important to you and when the time is right, makes a real-world commitment on your behalf. Sometimes all you need is a push.

• Gemini in Wear OS - Gemini is now on Wear OS smartwatches. Get a powerful AI assistant on your wrist to have natural voice conversations, manage tasks, and get information without pulling out your phone.

• Todo2 - Become a 100x Developer with AI-Powered Project Management Todo2 transforms developers into AI-powered project managers who never leave Cursor. Research, plan, and execute entire projects with superhuman productivity through MCP integration.

This week in AI

• YouTube Cracks Down on AI Slop - YouTube will demonetize mass-produced, repetitive, and low-effort AI-generated videos starting July 15, requiring original, authentic content for payouts.

• Isomorphic Labs Nears First AI Drug Trials - Alphabet’s Isomorphic Labs, spun out of DeepMind, is preparing its first human trials of AI-designed drugs for cancer, aiming to speed up and improve drug discovery using AlphaFold.

• AWS Unveils AI Agent Marketplace - AWS will launch an AI agent marketplace on July 15 with Anthropic as a key partner, letting startups sell AI agents directly to AWS customers.

• Gemma 3n MatFormer: Slice Custom Models - Gemma 3n’s MatFormer Lab lets you flexibly slice nested transformer models to match deployment needs, exporting custom sizes to Hugging Face with ease.

Paper of The Day

Enhanced Legal AI Framework combines 3-stage prompt engineering with multidimensional knowledge graphs for superior legal dispute analysis. Achieves 11.3% better sensitivity, 12.5% higher precision vs baselines. Features hierarchical prompts, multi-layer knowledge architecture & real-time legal updates.

📖 To read the whole paper visit here 👉 Link