Picture this: an innovative AI tool that lets you browse the web and even make purchases with just a few instructions – a real game-changer for busy shoppers everywhere. But what happens when a tech behemoth like Amazon decides to pull the plug on such advancements? That's the heated battle unfolding between Amazon and Perplexity, and it's stirring up questions about the future of AI in our daily lives. Stick around, because this clash isn't just about code; it's about who controls how we interact with the digital world.
The drama kicked off when Amazon.com fired off a cease and desist letter to Perplexity, demanding that the AI firm prevent its Comet browser from executing automated purchases on behalf of customers. Comet, much like OpenAI's Atlas and other cutting-edge browsers, harnesses a large language model to navigate the internet autonomously. This means it can handle tasks like surfing websites or completing online transactions at the user's command – essentially acting as a digital sidekick for e-commerce.
Amazon's stance is firm: they argue that any third-party apps with the power to shop for their users should get explicit permission from the e-commerce titan first. This, they claim, is crucial to safeguarding a top-notch customer experience. In a public statement, Amazon pointed out that tools like Perplexity's Comet carry the same responsibilities, and they've repeatedly urged Perplexity to exclude Amazon from Comet's features. Their reasoning? The bot's involvement leads to a noticeably poorer shopping journey and customer support. You can read their full take on it here (https://www.aboutamazon.com/news/company-news/amazon-perplexity-comet-statement).
From what we've gathered, Perplexity's Comet employs tactics to sidestep detection while engaging with Amazon's online marketplace. This echoes allegations in Reddit's lawsuit against Perplexity (https://www.theregister.com/2025/10/22/redditsuesperplexity/), where similar stealthy maneuvers are cited. Another worry for Amazon is that Comet's AI might opt for products outside the giant's tailored recommendations, potentially disrupting their personalized shopping ecosystem.
But here's where it gets controversial: Amazon could theoretically open up its personalization data to Perplexity's Comet through a partnership, though that might come with a price tag for data integration. Plus, Amazon seems keen on promoting its own in-house AI assistant, Rufus (https://www.aboutamazon.com/news/retail/amazon-rufus). Interestingly, Amazon's founder, Jeff Bezos, actually poured investment into Perplexity last year via his Bezos Expeditions Fund (https://www.perplexity.ai/hub/blog/perplexity-raises-series-b-funding-round). Talk about mixed signals!
Perplexity isn't backing down. They've hit back with a detailed blog post (https://www.perplexity.ai/hub/blog/bullying-is-not-innovation), framing Amazon's move as an attack on all internet users' freedoms. That might sound dramatic, but let's dive deeper into the data: a February 2025 survey by Omnisend, polling 1,026 US consumers (https://www.omnisend.com/blog/ai-trust-and-data-privacy-ecommerce/), revealed that 66% of people won't let AI handle purchases, even for potential bargains. Still, automated online shopping could be gaining traction as tech evolves.
Regardless, Perplexity labels Amazon's demands as outright bullying. Their legal rebuttal, which reads almost like it could have been penned by an AI itself, paints a vivid picture: 'For the past five decades, software has functioned as a tool, akin to a wrench wielded by its user. Yet, with agentic AI on the rise, software now embodies labor too – think of it as an assistant, employee, or agent.' They argue that big corporations like Amazon shouldn't have the power to veto your choice of digital helpers. 'This isn't sound legal reasoning,' they say, 'but a scare tactic to intimidate trailblazers like us from enhancing everyday life.'
To clarify for those new to this, agentic AI refers to systems that act autonomously on your behalf, going beyond simple tools to perform complex tasks like browsing or buying – it's like having a virtual personal shopper. Perplexity insists these AI agents differ from basic crawlers, scrapers, or bots, though they don't fully spell out the distinctions. They further contend that publishers and companies can't discriminate against users based on their AI preferences; everyone deserves the freedom to pick technologies that advocate for them.
And this is the part most people miss: while your trusty wrenches are safe from legal debates, software ventures into trickier territory. Experts are divided on whether AI agents stand apart from traditional web scrapers. Legal scholars at places like Santa Clara University, including professor Eric Goldman, have weighed in for The Register. Goldman notes that Comet uses Amazon customers' login details to facilitate buys, and platforms can indeed ban sharing those credentials in their terms of service. 'Any service has the option to restrict how users share their info,' he explains. 'They might stipulate in their agreements that credentials can't be passed to third parties.'
However, Goldman points out the flip side: such restrictions could alienate users who appreciate proxy services. For instance, financial apps often collect bank logins to pull data or authorize transactions for convenience. Banks could block this, but doing so might lose a valuable user segment. If Amazon enforces credential-sharing bans, users sharing with Comet could breach terms, and Perplexity might face liability for handling unauthorized info.
Goldman suggests Amazon could technically ban Comet outright if they can reliably spot it – though that's easier said than done, as evidenced by struggles to block AI crawlers (https://www.cjr.org/analysis/how-ai-browsers-sneak-past-blockers-and-paywalls.php). This would be a technical workaround, not a terms-of-service issue. He also speculates on potential lawsuits, like computer fraud claims, but emphasizes the legal uncertainty, especially post the US Supreme Court's Van Buren v. United States ruling (https://www.supremecourt.gov/opinions/20pdf/19-783_k53l.pdf) [PDF], which has left scraping laws in turmoil.
Goldman highlights that Comet's actions might not equate to standard web scraping, making it hard for laws to differentiate between human-like browsing, data harvesting, and AI-driven interactions. He draws a parallel to historic cable TV feuds, where networks and providers clashed over content distribution. 'Both sides might need to take some hits before they see the benefits of collaboration,' he says. 'That's how those carriage battles played out – eventually, money talks, and negotiations begin.'
We reached out to OpenAI and Anthropic for comments on whether they've faced similar cease-and-desist notices from Amazon regarding AI browsing, but no responses yet. Perplexity hasn't clarified if they'll comply with Amazon's demands.
This standoff raises intriguing debates: Is agentic AI simply an extension of user tools, or does it cross into uncharted legal waters as 'digital labor'? Should giants like Amazon dictate which AIs can shop on their sites, or should innovation reign free? And could this lead to a cooperative future, or more tech turf wars? What do you think: Are we on the brink of empowering AI assistants everywhere, or is Amazon right to guard its digital domain? Drop your opinions in the comments – let's discuss!
For more on AI's wild side, check out these related stories:
- Google envisions AI soaring in space via orbital data centers (https://www.theregister.com/2025/11/04/googletakesaiaspirationsorbital/)
- UK's Getty AI case fizzles out with no groundbreaking rulings (https://www.theregister.com/2025/11/04/ukcourtgettystabilityai/)
- Coders teaming with bots speed up work but cut corners (https://www.theregister.com/2025/11/04/humanhumancodepairsmoretalkative/)
- MIT Sloan pulls an AI ransomware study amid backlash (https://www.theregister.com/2025/11/03/mitsloanupdatesairansomware_paper/)
)
