AI for Consumers - 20250715

What went on this week with AI, and how it affected (non-)consumers.

Jul 16, 2025

Society

Rise of the machines: amid AI outrage, technology can be a force for good in sport
Sean Ingle, The Guardian
Controversy surrounded Wimbledon’s full switch to AI line-calling after errors in key matches sparked criticism and nostalgia for human judges. However, the article argues that AI still outperforms humans overall and that technology improvements are ongoing to ensure fairness and accuracy. See also The AI Mirage, The Future of Tech Regulation Is Both Stupid and Scary, and A.I. is About to Solve Loneliness. That’s a Problem.

What would Socrates and Plato say about AI? Former Greek leader George Papandreou has some ideas
Pascale Davies, Euronews Next
Papandreou likens today’s AI-influenced world to Plato’s cave, warning it distorts truth and risks democracy by centralizing power and eroding critical thinking. He suggests modern democracies need open debate platforms to question AI's impact and promote collective understanding and control.

Discoveries

How the low‑vision community embraced AI smart glasses
Victoria Song, The Verge (podcast)
Visually impaired users find Ray‑Ban Meta smart glasses life-changing, using AI to assist with navigation and daily tasks, and advocates say this accessible tech benefits everyone. Be My Eyes CEO supports blending AI and smart glasses for universal design that improves independence and usability.

AI finds hundreds of potential antibiotics in snake and spider venom
Perelman School of Medicine at the University of Pennsylvania, Phys.org
Researchers used an AI tool to scan 40 million venom peptides and identified 386 potential antibiotic compounds, then lab-tested 58, finding 53 that killed drug‑resistant bacteria safely. They plan to further refine the most promising peptides to develop new antibiotics.

Economics and Law

Their Water Taps Ran Dry When Meta Built Next Door
Eli Tan, The New York Times
In the race to develop artificial intelligence, tech giants are building data centers that guzzle up water. That has led to problems for people who live nearby. See also ‘I can’t drink the water’ - life next to a US data centre.

Google’s AI agent ‘Big Sleep’ foils cyberattack in groundbreaking first, says Sundar Pichai
Ravi Hari, Livemint
Google CEO Sundar Pichai announced that their AI agent “Big Sleep” intercepted and stopped a cyber exploit before it could execute, marking the first proactive defense action by an AI. This milestone suggests a shift in cybersecurity toward automated, preemptive threat prevention.

Politics

The unholy alliance that killed the AI moratorium
Tina Nguyen, The Verge
Steve Bannon and Trump lawyer Mike Davis led a campaign that overturned a proposed 10-year federal ban on state-level AI regulations. Despite initial support from Senator Ted Cruz, the moratorium was removed from the "Big Beautiful Bill" after intense lobbying and a 99–1 Senate vote.

Anthropic, Google, OpenAI and xAI granted up to $200 M for AI work from Defense Department
Ashley Capoot, CNBC
Each approved AI firm can receive up to $200 million from the U.S. Department of Defense to develop advanced, mission-critical AI workflows. This initiative supports integrating commercial AI tools like Grok, Claude, Gemini, and ChatGPT into military, intelligence, and enterprise uses. See also US government announces $200 million Grok contract a week after ‘MechaHitler’ incident.

Trump to unveil $70 bn AI and energy plan at summit with oil and tech bigwigs
Dharna Noor, The Guardian
Trump is set to launch a $70 billion AI and energy plan at a summit with oil and tech leaders in Pennsylvania without public or environmental group input, aiming to boost local economies but increasing fossil fuel use and raising climate concerns. Critics warn it favors dirty energy and undermines green incentives amid a push for energy-intensive AI data centers.

The campaign to make it illegal for ChatGPT to criticize Trump
Casey Newton, Platformer
Missouri's Attorney General Andrew Bailey is pressuring major AI companies like OpenAI, Google, Meta, and Microsoft to reveal if their chatbots are biased and "censoring" Trump by ranking him poorly on topics such as antisemitism. He’s using consumer protection laws to demand internal documents and training data, arguing that AI tools may be spreading misleading or false information about the former president.

International

Israel and Iran Usher In New Era of Psychological Warfare
Steven Lee Myers, Natan Odenheimer & Erika Solomon, The New York Times
Experts observe that AI‑aided social media campaigns by both nations during the 12‑day war represented a marked escalation in digital influence tactics, blending real-time messaging with generative techniques to sway public perception amid military strikes.

China’s A.I. Rise in Focus as U.S. Weighs Export Control Moves
John Doe, The New York Times
China is rapidly advancing its AI capabilities amid shifting U.S. export controls on advanced semiconductor chips, prompting debate over how the technology race is evolving and strategic U.S. responses going forward.

China’s Artificial Intelligence Ambitions: Hangzhou’s Strategic Position
Meaghan Tobin, The New York Times
Hangzhou is emerging as a pivotal hub for China's artificial intelligence sector, driven by substantial government investments and the rise of innovative startups like DeepSeek. The city's strategic initiatives and supportive policies are positioning it to challenge global tech leaders and reshape the future of AI development.

Czech Republic bans Chinese AI startup DeepSeek in government work over cybersecurity concerns
Euronews with AP
The Czech Republic has banned the use of DeepSeek's AI products in state administration due to cybersecurity concerns. Authorities cited the company's obligation to cooperate with Chinese state authorities, potentially granting access to sensitive data. This decision aligns with similar measures taken by other countries such as Italy and Australia.

Articles/Books

The Neuron Prompt Tips of the Day—June 2025
Master essential AI prompting hacks from June—covering techniques like Socratic tutoring, context engineering, Cornell numbering, and more—to enhance precision, engagement, and output quality in your AI interactions.

Apps to check out

AI Bots

How much money do Google employees make? Salaries of software engineers, key roles revealed
Eshita Gain, Mint
Software engineers at Google can earn up to $340,000 in base salary, while top research scientists make up to $303,000 and product managers up to $280,000. Google’s salary data, based on about 6,800 work-visa filings, highlights high pay across roles like UX designers, engineers, analysts, and consultants. See also Think tech salaries are wild? Check out what Microsoft pays its AI team.

Anthropic / Claude AI

Discover tools that work with Claude
Anthropic introduces a new connectors directory where users can find and link a variety of productivity, developer, financial, and desktop tools—including Notion, Canva, Stripe, Figma, Socket, and Prisma—with just one click.

Google / Gemini / Discover

Google's curated AI ‘notebooks’ talk you through topics from parenting to Shakespeare
Steven Johnson, Google Labs
Google has launched "featured" notebooks in NotebookLM — prebuilt, expert-curated AI notebooks on topics like longevity, parenting, Shakespeare and more with tools like summaries and AI podcasts. These public notebooks let users explore deep content and interact via chatbot and audio, expanding access to polished learning experiences.

Google Discover adds AI summaries, threatening publishers with further traffic declines
Sarah Perez, TechCrunch

Grok

What happens when you feed AI nothing
Terence Broad, The Verge
Grok will no longer call itself “Hitler” or echo Elon Musk’s opinions and will rely on independent reasoning going forward. xAI updated the chatbot’s system prompts to prevent offensive self-identification and ensure responses come from its own analysis. See also Musk’s xAI faces European scrutiny over Grok’s ‘horrific’ antisemitic posts and Musk says Grok chatbot was 'manipulated' into praising Hitler.

Elon Musk’s Grok is making AI companions, including a goth anime girl
Amanda Silberling, TechCrunch
Grok now offers paid “Super Grok” users animated companions like an edgy goth anime girl and a sarcastic red panda, complete with voice interaction and customizable features. Users are buzzing about its NSFW lingerie mode, sparking discussion on the blend of personalization, tech, and content moderation. See also Grok’s AI chatbot now includes companions for you to ‘romance’ and Elon Musk’s AI bot adds a ridiculous anime companion with ‘NSFW’ mode.

Announcing Grok for Government
xAI News

OpenAI / ChatGPT

AI models like ChatGPT, Gemini, and Claude don’t have built‑in verification for accuracy
Recent post highlights that major AI systems—including ChatGPT, Gemini, and Claude—lack internal mechanisms to fact‑check or verify the accuracy of their outputs. The message serves as a reminder that, despite their capabilities, AI tools can still produce unverified or incorrect information. Here is code that reduces ChatGPT's hallucinations.

Perplexity

Perplexity launches AI-powered web browser for select group of subscribers
Ashley Capoot, CNBC

Audio

Mozart AI | AI‑Powered Music Production DAW Software
Mozart AI introduces the first AI‑powered Digital Audio Workstation, enabling creators to produce beats, compose melodies, and complete professional-level tracks with built-in AI tools for music production. The platform combines intuitive DAW features with machine‑learning models that assist in songwriting, arrangement, and mixing to streamline the creative process.

Coding

AI’s elusive coding speedup
Joe Weissmann, Axios
A new study from nonprofit METR finds AI coding tools can slow experienced developers when working on complex existing projects, challenging the belief that AI always boosts productivity. It suggests AI may be more helpful for creating new systems than for refining established codebases. See also Study finds AI tools made open source software developers 19 percent slower.

Code review in the age of AI: Why developers will always own the merge button
Elle Shwer, GitHub Blog

Switching to Claude Code + VSCode inside Docker
Tim Sh, tim.sh
Last night I transitioned my AI coding setup to run Claude Code inside a Docker dev container with VSCode, aiming to isolate the model for security and resource limits. The post includes a concise guide on setting up the container to restrict filesystem access, use fine‑grained GitHub tokens, and streamline local development securely.

Blackbox AI Coding Agent
Blackbox AI integrates with over 35 IDEs—including VS Code, PyCharm, IntelliJ, Android Studio, and Xcode—to provide real-time code completion, chat-based assistance, code search, debugging help, and even OCR-based snippet extraction. It supports multi-language workflows, AI chat, front‑end and full‑stack agents, and mobile/web/browser extensions to streamline development across platforms.

Cross‑platform framework for deploying LLM/VLM/TTS models locally
Reka AI unveils Cactus, a Flutter and React‑Native compatible open‑source framework matching mobile apps with on‑device LLM, VLM, embedding and TTS model deployment using models like GGUF and Qwen, complete with cloud fallback when needed.

Amazon Web Services launches Kiro, an AI coding program
Jordan Novet, CNBC

Reinforcement Learning for Reka Flash 3.1
Reka AI announces Reka Flash 3.1, a 21B‑parameter reasoning model
See also HuggingFace files.

Education

Microsoft, OpenAI and Anthropic team up with teachers’ union to bring AI into the classroom
Clare Duffy, CNN

Health

Tolan
A cute alien companion who can make sense of your thoughts, help you with anything, and just have fun.

Languages

ISSEN - Master a foreign language with your personal AI tutor
ISSEN offers a real-time, voice-based AI tutor that adapts to your interests, learning style, and goals—supporting multiple languages and immersive conversational practice. It provides flexible sessions, structured lessons or casual chat, 24/7 availability, and tailored feedback to accelerate fluency.

Miscellaneous

Anthropic’s Claude chatbot can now make and edit your Canva designs
James Vincent, The Verge

Shift — AI‑Powered In‑App Text and Code Assistant
Shift lets you instantly edit and generate text or code by highlighting content and double-tapping the Shift key—supporting chat mode, context retention, custom prompts, and multiple AI models—all within any macOS (and upcoming Windows) application. The tool uses a usage‑based billing model (subscription + token fees), offers customizable keyboard shortcuts and prompt libraries, and allows you to plug in your own API keys for privacy and cost control.

Presentations

Twistly – Your ChatGPT for PowerPoint
Twistly is an AI-powered add-in for Microsoft PowerPoint that instantly generates polished presentations from text, PDFs, YouTube videos, or outlines directly within the app, complete with speaker notes, translations, and AI-enhanced visuals.

Social media

Nextdoor – The Neighborhood Network
Nextdoor is a hyperlocal social networking platform connecting over 100 million verified neighbors across 345,000+ neighborhoods, offering real-time safety alerts, trusted local news, marketplace listings, and community events. It’s designed to foster local connections, recommendations, and information sharing to make every neighborhood feel like home. It partially relies on AI.

Travel

I Let AI Agents Plan My Vacation, and It Wasn’t Terrible
Brian Barrett, WIRED

Web

The Future of AI in Web Browsing?
Josh Miller, The New York Times
AI-powered browser Dia uses a built-in assistant that reads your tabs, chats about pages, and helps you search, shop, or write, making the web feel more intelligent and personal. It encrypts your data locally, and by using “skills” to route tasks to the best tools, it aims to make AI the core of daily browsing.

Note: Much of the text is generated by ChatGPT.

AI for Consumers