Good morning, AI enthusiast.

Anthropic is making a significant change to its privacy policy, announcing that conversations from its consumer Claude users will now be used to train future AI models through a new opt-out system.

The move highlights the intense demand for real-world user data needed to stay competitive in the AI race. But does this shift place too much of a burden on users to be vigilant and actively manage their own data privacy settings?

In today’s AI recap:

• How We Generate Unlimited Ad Creative with Nano Banana + n8n (Gemini 2.5 Flash Image)

• Anthropic to train Claude on user chats

• A new study on AI’s threat to entry-level jobs

• Taco Bell’s AI drive-thru experiment struggles

• Cloudflare’s one-click conversational websites

Claude's Privacy Shift

The Recap: Anthropic is reversing its policy and will now use conversations from its free and paid consumer users to train future AI models, shifting to an opt-out system.

Unpacked:

• The new policy applies to Claude Free, Pro, and Max plans, but does not affect enterprise accounts like Claude for Work, Claude Gov, or users accessing the model via API.

• Existing users have until September 28, 2025, to make their selection, after which a choice is required to continue using the service; new users will see the option during signup.

• If you allow your data to be used, Anthropic is extending its data retention period from 30 days to five years, though you can change your choice in your Privacy Settings at any time.

Bottom line: This move highlights the intense demand for real-world user data to stay competitive in the AI race. It also shifts the responsibility to users to be vigilant and actively manage their data privacy settings.

Is Your Amazon Strategy Actually Working? Here’s What Top Brands Do Differently.

At Cartograph, we’ve worked with some of the most innovative brands in CPG—OLIPOP, Starface, and Rao’s—and understand the nuances of selling consumables on Amazon.

Are you a fast-growing brand in Food & Beverage, Supplements, Beauty & Personal Care, Household, Pet, or Baby?
Growing 50%+ YoY?
Do you know your Amazon profitability (and are you happy with it)?

We’ve spent the past 7 years helping CPG brands scale profitably on Amazon. What makes Cartograph different:
• Deep CPG focus
• No more than 4 brands per team
• Monthly P&L forecasts within 5% accuracy
• Daily reporting via Slack

Click below to get a custom, human review of your Amazon account—not just another automated report.

Get An Amazon Audit

AI's Entry-Level Threat

The Recap:** A new Stanford study using ADP payroll data reveals that jobs for young workers most exposed to AI have seen a 13% relative decline since 2022. The findings offer some of the first large-scale evidence of AI's direct impact on the labor market.

Unpacked:

• The decline specifically affects workers aged 22-25 in roles like software development, accounting, and customer service.

• Researchers suggest AI is more capable of replacing the codified knowledge from formal education than the tacit knowledge gained from years of experience.

• Job losses are concentrated where AI automates tasks, while employment remains steady in roles where AI augments and assists human workers.

Bottom line: This study moves the conversation from theoretical to tangible, showing AI's immediate effect on entry-level hiring. For young professionals, the path forward involves mastering skills that leverage AI tools rather than competing against them.

AI Training

How I Generate Unlimited Ad Creative with Nano Banana + n8n (Gemini 2.5 Flash Image)

The Recap: In this video, I'm going to show you how to set up and use Nano Banana inside of an n8n automation (Google's brand new AI image generator and image editor Gemini 2.5 Flash Image).

We're going to build an AI automation that generates high-quality ads at scale by putting a real product into the hands of many different influencers.

P.S. We also launched a free AI Automation Community for those looking to build and sell AI Automations — Come join us!

Taco Bell's AI Bust

The Recap: Taco Bell is reevaluating its AI-powered drive-thru experiment after the system struggled with order accuracy and became a target for customer pranks.

Unpacked:

• Deployed in over 500 restaurants, the AI has been trolled with absurd orders, including one for 18,000 cups of water, highlighting a major human-computer interaction challenge.

• This isn't an isolated case, as other chains like McDonald's and Wendy's are also navigating the complexities of implementing drive-thru AI with mixed results.

• The company is now considering a hybrid approach, where humans might take over during busy periods when the AI system falters under pressure.

Bottom line: This setback shows that deploying AI in real-world, consumer-facing roles is more complex than just getting the technology to work. Success depends just as much on understanding customer behavior and designing for unpredictable human interaction.

Where AI Experts Share Their Best Work

Join our Free AI Automation Community

Join our FREE community AI Automation Mastery — where entrepreneurs, AI builders, and AI agency owners share templates, solve problems together, and learn from each other's wins (and mistakes).

What makes our community different:

• Real peer support from people building actual AI businesses

• Complete access to download our automation library of battle-tested n8n templates

• Collaborate and problem-solve with AI experts when you get stuck

Dive into our course materials, collaborate with experienced builders, and turn automation challenges into shared wins. Join here (completely free).

Cloudflare's Conversational Web

The Recap: Cloudflare and Microsoft have launched AutoRAG and NLWeb, a new one-click solution that crawls any website to instantly make it conversational for both users and AI agents.

Unpacked:

• The system works by using AutoRAG to automatically crawl a site, store its content, and embed it into a managed vector database for semantic search.

• It’s all built on NLWeb, an open project from Microsoft that defines a standard protocol for natural language queries on websites.

• This approach prepares websites for the agentic web, creating a structured way for AI agents to retrieve information accurately and move beyond keyword search.

Bottom line: This move lowers the barrier for publishers to create on-site answer engines without managing complex infrastructure. It also signals a shift toward a more structured, AI-friendly web where sites can serve information directly to agents on their own terms.

The Shortlist

OpenAI updated its Model Spec based on global public surveys, incorporating feedback on model behavior and releasing the dataset of over 1,000 participants' preferences.

Maisa AI raised a $25M seed round to address the 95% failure rate of enterprise AI pilots by offering a platform for deploying accountable and trainable "digital workers."

Wiz detailed a major npm supply chain attack where malicious packages weaponized installed AI CLI tools to systematically harvest thousands of enterprise developer credentials.

Dedalus Labs launched its YC S25-backed "Vercel for Agents," a cloud platform designed to simplify building agentic AI applications by connecting LLMs to MCP tools with a single API.

Signing off,

David, Lucas, Mitchell — The Recap editorial team