AI Agents: Blockages, Amazon Lawsuits, Shopify Files

Some platforms are closing the door to AI agents while others are already welcoming them. Between Amazon's call in its litigation with Perplexity and dedicated payment queues at Shopify, the access of agents to the web is now at stake. Their rapid proliferation, illustrated by Meta's Muse and the increase in agent traffic, precedes rules that are yet to be determined.

Debated Rules Will Decide the Web of Agents

The rules governing the interaction between AI agents and websites are still being developed. Some, such as those aimed at preventing the exposure of credit card information during automated journeys, are considered beneficial for all. Other aspects are controversial, particularly the ability for companies to block agents' access or prohibit them from negotiating better deals for consumers. The direction of these decisions will determine whether the web becomes a space where agents act in the interest of users or not.

Blockages, Lawsuits, and Advertising Interests at Stake

Some platforms, particularly those focused on security, prohibit access to AI agents. Difficulties have been reported in using these agents on sites like Pizza Hut and Yelp, the latter refusing table reservations made by AI. Amazon has blocked Muse agents and is suing Perplexity to prevent its agents from operating. A judge recently ruled in favor of Perplexity, but Amazon has appealed, arguing that AI agents complicate the distinction between human views and automated traffic, which would threaten its advertising business. Additionally, many sites block agents sent by individuals to negotiate prices or organize trips.

E-commerce Opens Dedicated Paths for Agents

The majority of web traffic generated by AI agents pertains to e-commerce. Platforms in this sector have a financial interest in allowing these agents to make purchases on behalf of real people. Shopify is already welcoming AI agents by providing them with specific payment queues that bypass pages intended for humans.

Accelerated Adoption: Muse Reaches 5 Million Downloads

The adoption of personal AI agents is progressing rapidly, with massive launches via applications such as Dots from OpenAI and Muse from Meta. Muse recorded 5 million downloads in its first month, while the volume of web traffic associated with these agents tripled between August and September. These solutions enable users to accomplish online tasks, such as booking a flight or hotel.

Possible Consequences for Humans and Conditional Promises

AI agents form a third category of users, distinct from humans and bots. Some scenarios envision a web that is less welcoming to people: degraded interfaces, repeated humanity checks that could include iris scans, variable pricing depending on the agent used, and recommendations influenced by advertisements or platform commissions. However, Google, Anthropic, and OpenAI promise that AI agents will empower consumers to negotiate better, for example, to contest medical bills or find cheaper car insurance. Ethan Mollick, a professor at the Wharton School, anticipates that widespread use of agents to obtain refunds or negotiate prices could overwhelm customer service and irritate users if their agents are denied these requests.