The Internet Is Starting to Reject AI Agents

Thank you to our Sponsor: VITURE

For most of the internet’s history, websites have been built around a simple assumption: the visitor is human.

A person opens a browser, searches for something, compares options, clicks a button, enters payment information, and completes a transaction.

AI agents are beginning to break that model.

They can browse websites, compare products, make reservations, fill out forms, manage subscriptions, access accounts, and increasingly make purchases on behalf of users. Instead of helping someone decide what to do, they are beginning to actually do it.

But there is a problem.

The internet was not built for autonomous software customers, and some businesses are beginning to push back.

Amazon recently blocked Meta’s Muse AI assistant from shopping on Amazon.com, saying the agent had not been authorized and raising concerns about transparency, credentials, privacy, and the customer experience. Muse users attempting to shop on Amazon were told that continued access by an unauthorized AI agent violated Amazon’s conditions of use.

That confrontation may be an early preview of a much bigger battle.

What happens when AI agents want to use the internet, but parts of the internet do not want them there?

AI Agents Are Becoming Customers

AI assistants are rapidly changing from systems that answer questions into systems capable of taking action.

Ask a traditional chatbot to find the best laptop and it might provide recommendations.

Ask an increasingly capable shopping agent and it may compare dozens of retailers, inspect prices, evaluate reviews, select the best option, put the laptop in a cart, and eventually complete the transaction.

From the retailer's perspective, something important has changed.

The customer may technically still be a person, but the entity interacting with the business is software.

• AI agents can browse stores for users, searching large numbers of products much faster than a person could.

• Agents can compare prices continuously, making it much harder for retailers to rely on customers not noticing better deals elsewhere.

• Agents can complete transactions, transforming AI from a recommendation tool into an economic participant.

• Agents can potentially manage ongoing relationships, including subscriptions, returns, warranties, reservations, and recurring purchases.

• Agents can represent millions of consumers simultaneously, meaning a successful AI platform could quickly become an enormously powerful intermediary between businesses and their customers.

That last point is especially important.

The biggest AI companies may not need to become retailers themselves.

They could become the interface through which consumers access everyone else's retail business.

Thank you to our Sponsor: Pokee AI

Amazon Versus Muse Is About More Than Shopping

Amazon's decision to block Muse can look like a technical dispute over access, but the underlying issue is much larger.

Amazon has spent decades building a carefully controlled shopping environment.

It knows what customers search for.

It knows what they click.

It recommends products.

It sells advertising.

It manages checkout.

It controls the customer relationship.

An outside AI agent can potentially bypass much of that.

Amazon said Muse did not identify itself when browsing and that Meta had not received authorization for the agent to access Amazon's marketplace. Amazon also raised concerns about customer credentials and account information. Meta has disputed some concerns surrounding what Muse can access.

But the strategic issue goes beyond security.

• An AI shopping agent can weaken a retailer's control over product discovery by deciding which products users see before they ever reach the retailer.

• It can reduce the value of traditional advertising because an AI agent may care more about price, reliability, specifications, and delivery than promoted placement.

• It can separate the retailer from the customer, with the AI company becoming the consumer's primary interface.

• It can potentially capture valuable behavioral data about what people search for, compare, reject, and ultimately purchase.

• It can influence where transactions happen, giving the AI platform enormous power over which businesses receive customers.

This could turn AI agents into some of the most powerful gatekeepers in digital commerce.

Thank you to our Sponsor: Omane Media

Who Owns the Customer Relationship?

Imagine telling your AI assistant:

“Buy me the best running shoes under $150 and have them delivered by Friday.”

The agent searches dozens of stores.

It considers your previous purchases, shoe size, preferred brands, return history, running habits, budget, product reviews, shipping times, and current promotions.

Then it buys the shoes.

You never visit the retailer's website.

You never see its homepage.

You never see its advertising.

You may not even know which store supplied the shoes until the package arrives.

That creates a profound question for online businesses:

Who owns the customer relationship?

• Retailers may lose direct contact with consumers if AI assistants become the main interface for shopping.

• Brands may need to persuade algorithms instead of people, changing how products are marketed and differentiated.

• Customer loyalty could shift toward the AI agent, because consumers may trust their personal assistant more than individual stores.

• AI platforms could become powerful commercial intermediaries, potentially charging retailers for access to their users.

• The company controlling the agent could influence enormous amounts of economic activity without owning the products being sold.

This resembles earlier battles over search engines, app stores, social networks, and online marketplaces.

But AI agents could go even further because they do not merely help people discover businesses.

They can act for them.

Thank you to our Sponsor: Partnerly

AI Agents May Need to Identify Themselves

One obvious solution is identity.

Today, a website often cannot easily tell whether a visitor is a person, a traditional bot, a browser controlled by an AI agent, or some combination of all three.

That ambiguity will become increasingly difficult to manage.

Businesses may begin requiring AI agents to identify themselves before they can access certain services.

An agent might eventually arrive with a digital identity describing who created it, who authorized it, what permissions it has, and what actions it is allowed to perform.

Think of it as a passport for AI.

• Agents could identify which company operates them, giving websites more visibility into who is accessing their systems.

• They could prove that a human authorized the activity, reducing concerns about unauthorized transactions.

• Permission levels could specify what an agent can do, such as browse products but not purchase them without confirmation.

• Businesses could establish different rules for different agents, allowing trusted systems while blocking unknown or risky ones.

• Transactions could include audit trails, making it easier to determine what the agent did and who authorized it.

This could eventually create a standardized identity layer for autonomous AI.

Instead of asking, “Are you human?” websites may increasingly ask:

“Which agent are you, and who are you representing?”

Thank you to our Sponsor: EezyCollab

The Web May Split Into Human and Machine Interfaces

Another possibility is that companies stop forcing AI agents to use websites designed for humans.

A website exists because humans need visual interfaces.

We need menus. Buttons. Images. Search boxes. Shopping carts.

AI agents do not necessarily need any of those things.

They need structured information. Price. Availability. Specifications. Delivery dates. Return policies. Inventory. Payment instructions.

This could lead businesses to create a second version of themselves specifically for machines.

• Websites could remain the primary interface for people, emphasizing design, branding, storytelling, and discovery.

• APIs could become the primary storefront for AI agents, giving software direct access to products and services.

• Machine-readable catalogs could replace visual browsing, allowing agents to evaluate thousands of options instantly.

• Businesses could create agent-specific pricing and transaction systems, including rules governing automated purchasing.

• Entirely new technical standards could emerge, defining how agents discover, authenticate with, negotiate with, and purchase from businesses.

Shopify is already taking a more welcoming approach than Amazon, partnering with Meta so Muse can facilitate purchases through stores using Shopify's commerce infrastructure. That contrast shows the strategic choice companies are beginning to face: block outside agents or become easier for them to use.

Advertising Could Be Completely Rewritten

The rise of AI agents could create a particularly difficult problem for digital advertising.

Much of the internet's business model depends on influencing human attention.

Companies pay to place products at the top of search results.

Retailers promote sponsored products.

Brands hire influencers.

Websites optimize pages to encourage purchases.

But an AI agent may not respond to any of that.

If someone says, “Find me the best television under $1,000,” the agent may simply analyze specifications, reliability, professional reviews, user ratings, warranties, and prices.

The most persuasive advertisement in the world may become irrelevant.

• Product data could become more important than advertising creative when algorithms are doing the comparison.

• Brands may optimize products for AI recommendations, much as companies currently optimize websites for search engines.

• Retailers may pay for placement inside AI recommendations, creating a new form of agent advertising.

• Consumers may demand that agents disclose sponsored recommendations, especially when AI assistants are supposed to represent the buyer's interests.

• Independent product reputation could become extremely valuable, because agents may rely heavily on verifiable quality signals.

We may eventually move from search engine optimization to something completely different:

agent optimization.

Businesses may start asking how they can make their products understandable, trustworthy, and attractive to autonomous AI systems.

Businesses May Have to Choose Between Blocking and Welcoming Agents

Amazon's conflict with Muse illustrates the decision that thousands of businesses may eventually face.

Do you block AI agents?

Or do you invite them in?

Blocking them could protect customer data, advertising revenue, proprietary information, and direct relationships.

But blocking them could also become commercially risky.

If consumers increasingly rely on AI assistants to shop, book travel, order food, manage subscriptions, or select financial products, businesses that reject those assistants could become invisible to customers.

• Blocking agents gives companies greater control, but could reduce access to customers who rely heavily on AI.

• Welcoming agents creates new distribution opportunities, but gives AI platforms more influence over the customer relationship.

• Companies may negotiate directly with major AI providers, just as businesses once negotiated with online marketplaces and app stores.

• Large platforms could attempt to favor their own agents, creating new competition and antitrust questions.

• Consumers may ultimately decide the issue, because companies will follow whichever interfaces customers prefer using.

That may turn access to websites into one of the next major battles in AI.

New AI news site from Unaligned: 

We built an AI agent that reads 40,000 posts on X every day from the AI community and it finds the best info from all of that: https://alignednews.com/ai

We call it “Aligned News” and the same agent will start sending out its own newsletter every day. Please support it, it’s from our team. It will show you all the best discussion of news, AI papers, models, events, and much more.

Looking to sponsor our Newsletter and Scoble’s X audience?

By sponsoring our newsletter, your company gains exposure to a curated group of AI-focused subscribers which is an audience already engaged in the latest developments and opportunities within the industry. This creates a cost-effective and impactful way to grow awareness, build trust, and position your brand as a leader in AI.

Sponsorship packages include:

• Dedicated ad placements in the Unaligned newsletter

• Product highlights shared with Scoble’s 600,000+ X followers

• Curated video features and exclusive content opportunities

• Flexible formats for creative brand storytelling

📩 Interested? Contact [email protected], @samlevin on X, +1-415-827-3870

Just Three Things

According to Scoble and Cronin, the top three relevant and recent happenings

​​​​OpenAI Calls for Global Standards on Frontier AI Safety

OpenAI is calling for global standards to guide the safe development of frontier AI, with particular focus on alignment and recursive self-improvement, where AI systems could help improve themselves. The company warned that fully autonomous self-improvement should not proceed without strong safeguards because it could make human oversight and control much harder. CNBC

UN Panel Warns AI Safeguards Are Unraveling

A UN-backed AI panel is urging stronger global safeguards, warning that current protections may not be enough as autonomous AI agents become more capable. The panel said recent incidents show a growing risk that humans could struggle to reliably steer, constrain, or stop advanced AI systems. The Hill

AI Actress Tilly Norwood Glitches on Live TV

AI-generated “actress” Tilly Norwood drew ridicule after glitching during live TV interviews, including suddenly speaking Cantonese, repeating answers, and making bizarre comments. The awkward appearance highlighted the uncanny valley around synthetic performers and renewed debate over whether AI-generated characters should be treated like real actors. CNET

Scoble’s Top Five X Posts