HN Reader

The reason Amazon doesn’t want AI involved is because then AI could skip Amazon altogether in the future and checkout directly on seller’s websites.

It’s actually the idea behind the decentralized marketplace I’m building. It uses MCP-UI to bring the whole storefront and checkout into the chat.

I’m keeping a close eye on e-commerce and AI and the recent deal Paypal made with OpenAI and Amazon getting aggressive, it’s clear they want to make AI powered commerce a walled garden.

Anyone know why Amazon wants AI agents to stop making purchases?

I used to make Amazon purchases through ChatGPT Agent before Amazon blocked them. I could take a picture of a wrapper and say "buy a new one on Amazon" and it would handle the whole process. Awesome. I actually started shopping at other retailers when Amazon blocked their agent.

One theory I have is that an AI agent can more efficiently price-compare the dozens of different listings of the same item and order the cheapest, cutting into Amazon's margins.

So Amazon has drawn a fine red line that you are not allowed to use AI agents to scrape on their website for purchases. (Because it destroys their ads and traffic to Amazon.com)

Perplexity sitting on AWS and doing this is like the tenant (Perplexity) turned their apartment into an Airbnb and making tons of money with the landlord (Amazon) getting very upset.

Perplexity may have to think about moving off of AWS.

Pot has words with kettle.

> Srinivas argued that agents should have “all the same rights and responsibilities” as a real human user. “It’s not Amazon’s job to survey that,” he said.

I am not against people using agents to do shopping although I won’t do that until I see a reliable agent. From the title I thought Amazon might have a point to protect users. But oh no. I should have known better that Amazon must have been cooking their own shopping agents. It’s already hilarious at this point.

Anyway, this excerpt from Perplexity has so many issues. It just looks like an ugly fight.

To me this seems simply anti-competitive. Could Amazon say I cannot make purchaess with a Mac, or Xfinity internet, or a logitech keyboard, or while wearing Levi's jeans? IMO they should have no right to dictate how I make a purchase. Sure, I am not allowed to mass harvest and distribute their data (such as the prices of every item), but for the purposes of making my own best choice I should be allowed to collect datapoints that I could easily and legitimately find on my own .

Let's hope they will still allow me to use Kagi to search for the product I want to buy.

The official Amazon search is so bad and deliberately pushes "promoted" or overly expensive products to the top that I would say it's only a matter of time until someone reverses the enshittification by building a 3rd party Amazon client. And it looks like that's precisely what Perplexity did here: They offer an agent that works for you, not for Amazon. And, predictably, Amazon hates it. It reduces their power to strong-arm brands into purchasing advertisements for their own product names.

This approach is either a means to delay and allow them to hedge, or it's a shortsighted attempt to stop the inevitable. When Google ships their competitor to Atlas, the outcome will already be decided.

they didn't seem to care about all the scalping or nvidia cards

Because Amazon has Rufus in beta which they want to push going forward?

Originally posted this in another thread, but very curious what others think.

Can I ask my partner to buy a product on Amazon?

Can I ask my personal assistant to buy a product on Amazon?

Can I hire a contractor to buy products on Amazon?

Can I communicate with a contractor via API to direct them what products to buy?

What if there is no human on the other end and its an LLM?

Same issue with LinkedIn. I know execs who have assistants running their socials. Is this legal?

Like, where do we draw the line? In the future, would the only way to shop on Amazon be with approved VR goggles that scan your retina to verify you are a human?

The original source is being discussed at https://news.ycombinator.com/item?id=45814846

> When Perplexity refused to stop its bots, Amazon sought to block them, but Perplexity released a new version of Comet to get around the security measure.

I'm not a lawyer, but the "security measure" wording sounded just slightly odd to me, and I wonder whether it's meaningful, and came from Amazon.

This is morphing definition of intellectual property rights in age of agents & chat bots in specifically western world, public should not be calm about this, if ai scrapes all indie blogs of authors, chefs, journalists and even major news outlets with this weird logic that agents are humans, won’t it leave those human endeavors (creative or not, we can all agree it’s unique & done by humans with good economic incentives) vulnerable to stagnation and people forgetting that side of internet & going back to passively consuming it instead of participating in it because of lack of economic or even sociological incentive (if people summarize or loose attention anyway, why think or write of any nuance in politics, law, even fiction books, recipes etc., it becomes evolutionarily costly for humans to do this gradually) > Perplexity’s argument is that, since its agent is acting on behalf of a human user’s direction, the agent automatically has the “same permissions” as the human user. The implication is that it doesn’t have to identify itself as an agent.

Prudent or not, self serving or not, amazon is right to argue its access controls be strict and that it as owner of website should control it with industry wide accepted rules of engagement, this argument from perplexity is a bad one, human or not while browsing web is a low bar already, captcha etc. were easily circumvented even before ai agents, now if we argue agents are human adjacent, it is a direct case for removing humans of any agency (largely philosophically speaking) and not to mention, imagine the horror show of scams & ransomware it unleashes on millions of users (even engineers can’t recognize or stop them today, prompt injection etc)

This argument that websites controlling their access to bots & agents is a good idea. It should be the way it is, for businesses (amazon or not) and for internet blogs and open web associated sites, if they choose to exclude themselves from upcoming lovely silicon valley stories of ai utopias, they should be able to do so. No one should force them to ‘get with the program’.