OCTOBER 29, 2025

• PromptCraft: Ways to Make Prompts Better, Self-Evaluation, and 5 That Will Change the Way You Work
• NewsDive: Proposed Ban on Teens Using AI Is So Wrongheaded, It’s Embarrassing Adults Actually Thought of It​
• THE HEADLINES: Bill Gates' Thoughts on AI; OpenAI urges US to Increase Power-Generating Capacity to Compete with China; Chegg and Amazon Slash Thousands of Jobs; Hurricane Melissa Brings Out the Deepfakes; and Grokipedia Is Now a Thing
• Tool Spotlight: EraseVideo Removes Video Watermarks for All Major Models

Ways to Make Prompts Better, Self-Evaluation, and 5 That Will Change the Way You Work

• Self-evaluation prompt for AI output quality: A meta-prompt that scores a single response across seven failure modes (e.g., tunnel vision, over-assumption), asks for examples, impact, and improvements, then summarizes strengths/risks. A lightweight QA step to reduce hallucinations and fluff.
• Prompt-Alchemy guide to steer tone and depth: Short practice guide for “tuning” model behavior via perspective, rhythm, and framing. Offers repeatable patterns (analytical vs. reflective vs. creative cues) and an exercise to compare outputs across styles for better conversational control.
• Five prompt structures that consistently deliver: A compact set: role-playing expert, brainstorm-and-categorize, summarize-and-extract actions, simplify-and-explain with analogy, and condense-and-refine with “explain why” edits. Clear formats that trade vagueness for reproducible, actionable outputs.
• ArtisMind: structured prompt engineering workflow: Tool advertises requirements capture, structured prompt assembly, multi-model testing, and security checks, plus context file support. For teams moving from ad-hoc prompting to repeatable, testable prompt specs and quality scoring.
• Universal prompt to generate better prompts: A “prompt-for-prompts” assistant that gathers GOAL, CONTEXT, and FORMAT, adds examples, schemas, constraints, evaluation criteria, and an optional reasoning step—then offers an alternative version. Good scaffolding for fit-for-purpose prompts.

Proposed Ban on Teens Using AI Is So Wrongheaded, It’s Embarrassing Adults Actually Thought of It

On Tuesday morning, US Senators Josh Hawley and Richard Blumenthal rolled out the GUARD Act, a bill that would force AI companies to verify every user’s age (think government IDs or face scans), make chatbots announce “I’m not human” at the start of a conversation and every 30 minutes, and ban minors from using “AI companions.”

It also creates civil and criminal penalties for chatbots that encourage self-harm or sexual content with minors. The full text is out, and it’s sweeping.

I get why this is happening.

Parents and safety advocates have testified about heartbreaking cases. OpenAI has already said it will stop discussing suicide with teen users and has shared sobering stats about the volume of suicidal ideation it sees.

Those harms are real; the fear is understandable. But fear shouldn’t write policy.

Here’s the problem: this bill confuses surgical safety work with blanket prohibition, and it does so in a way that would make the internet less private, less equitable, and less educational for the very kids we’re trying to help.

What the bill would actually do

• Age-gate everyone. All chatbot users, adults included, would be forced through “reasonable” age verification. That typically means ID uploads, selfies, or third-party age estimation. That data has to live somewhere. They’ll be breached somewhere. And many adults (and plenty of teens in precarious circumstances) simply won’t provide them.
• Hard-ban minors from “AI companions.” The definition targets bots designed to simulate friendship or therapy. But in practice, cautious platforms will likely over-block teen access to tutoring and study helpers to avoid risk.
• Mandate repetitive disclosures. “I’m an AI, not a human” at the start and every 30 minutes. Imagine telling your smart speaker, “Turn on the lights,” and getting, “Okay, and quick reminder, I’m not a human!” again and again. That’s not safety; that’s friction theater.

CHECK OUT WHY IT'S SUCH A BAD IDEA ON SUBSTACK

HIGHLIGHTS: Bill Gates' Thoughts on AI; OpenAI urges US to Increase Power-Generating Capacity to Compete with China; Chegg and Amazon Slash Thousands of Jobs; Hurricane Melissa Brings Out the Deepfakes; and Grokipedia Is Now a Thing

1. Bill Gates: AI is the biggest technical thing ever in my lifetime
2. OpenAI calls on U.S. to build 100 gigawatts of additional power-generating capacity per year — says electricity is a ‘strategic asset’ in AI race against China
3. Hit by AI, edtech firm Chegg slashes jobs and names new CEO in major overhaul
4. PayPal Launches Agentic Commerce Services to Power AI-Driven Shopping
5. Amazon cuts 14,000 corporate jobs as spending on artificial intelligence accelerates
6. Fake AI videos clog social media as Hurricane Melissa bears down on Jamaica
7. Qualcomm is turning parts from cellphone chips into AI chips to rival Nvidia
8. Elon Musk launches Grokipedia to compete with online encyclopedia Wikipedia
9. Advancing Claude for Financial Services
10. AI chatbots use different sources than Google search and often cite less-known websites
11. Strengthening ChatGPT’s responses in sensitive conversations
12. Groq CEO: Middle East is the ‘ideal place’ for data centers
13. OpenAI is reportedly working on an AI music-generation tool
14. DeepSeek and Qwen AI models crush Western rivals in cryptocurrency trading challenge
15. AI can help authors beat writer’s block, says Bloomsbury chief
16. Exclusive: US Department of Energy forms $1 billion supercomputer and AI partnership with AMD

EraseVideo Removes Video Watermarks for All Major Models

EraseVideo is a free Mac app that removes visible watermarks from AI-generated videos (including Sora, Veo3, and Keling) in about a minute—running entirely offline for speed and privacy with no logins or uploads.

It’s built for simple workflows, helping creators quickly clean drafts for demos or internal reviews without cloud tools. Use responsibly and only for content you own or have permission to edit.

Learn More: https://erasevideo.app.

Read All mAIn Street Back Issues Here

Your input fuels the content. Let me know how I did today!