Amazon Web Services (AWS) is taking on the growing scourge of traffic from generative AI applications with a filtering tool for its Network Firewall.

Amid the rise of apps like Perplexity, ChatGPT, and Microsoft Copilot, a huge wave of AI-induced traffic has seen internet traffic soar, but often misconstruing actual user numbers. Such apps sift the internet looking for any relevant sources, taking on board a webpage’s contents before returning the information to a user via a regurgitated summary.

Content delivery platforms like Cloudflare have rallied against such traffic, going as far as to call out Perplexity for allegedly circumventing blocks put in place to stop AI crawlers from accessing its customers’ content for search responses without permission.

AWS is now looking to shore up platform access, with a new capability to its cloud-native firewall platform enabling users to control access to generative AI services, as well as social media platforms, and streaming sites directly within firewall rules via pre-defined URL categories.

Available in all AWS commercial regions where AWS Network Firewall is supported, the hyperscaler touts the feature as being able to restrict AI access “while reducing operational overhead.”

It’s also being marketed as a tool for helping meet regulatory requirements, such as with stipulations of the U.K.’s Online Safety Act, or the EU’s Digital Markets Act. AWS said its revised firewall offering "helps security and compliance teams enforce consistent policies" through more clearly defined inspections of traffic based on URL categories.

Amazon Web Services added a managed network firewall service to its security tool chest and says GE and the U.S. Navy are among the customers using AWS Network Firewall.

Introduced back in 2020, AWS’s managed network firewall service can run in the Amazon Virtual Private Cloud environment and is capable of scaling with network traffic to protect customers’ workloads.

The offering found itself dead last in rankings by nonprofit testing lab CyberRatings published in 2024, with an apparent effectiveness score of just 5.39%, with the likes of Versa Networks' NGFX and Check Point CloudGuard coming out at the top end.

The hyperscaler went on to refuse the nonprofit’s findings, contending that the report was “inaccurate and incomplete.”

CyberRatings 2025 cloud firewall findings, however, saw AWS receive a score of 0% security effectiveness. Hyperscale rivals Google Cloud and Microsoft Azure weren’t left unscathed, also achieving nil poi as "they allowed attacks to bypass existing defenses.”