Le Monde delivers sub-60ms personalization and protects high-value archives from AI bots at the edge
Key results
Achieved a record-breaking 58.98ms average response time during major election traffic spikes
Served 21 million visits and 70 million pageviews in a single day without origin scaling
Pivoted from traditional blacklisting to a strict whitelist security posture, blocking unapproved AI crawlers across 3 million archived articles
Blocked approximately 100 million attack requests monthly using Fastly Next-Gen WAF
Pioneered real-time edge personalization in Rust on Fastly Compute to stitch customized subscriber content into cached page templates
The challenge
Groupe Le Monde is one of France's premier news organizations, supported by nearly 900 journalists and more than 1.5 million subscribers across its portfolio. For a major digital news outlet, election nights represent the ultimate operational test. Traditionally, Le Monde routed nearly all traffic back to origin servers to deliver personalized page elements for subscribers. But during high-stakes breaking news events like these, audience surges threatened to overwhelm backend infrastructure and degrade response times.
Simultaneously, the rise of generative AI introduced a new operational threat: unauthorized content scraping. Le Monde maintains a vast digital archive of nearly 3 million articles written in high-quality French, spanning from the end of World War II to the present day. Because Large Language Model (LLM) developers view verified French archives as highly valuable training data, unvetted AI bots and third-party data aggregators constantly scraped the site. Le Monde needed a way to deliver sub-second page performance under massive load while enforcing strict monetization and licensing controls over its intellectual property.
The solution
Le Monde expanded its long-standing Fastly partnership, moving beyond global CDN delivery to adopt Fastly Compute, Next-Gen WAF, and AI Bot Management with ContentGuard.
Execute edge personalization in Rust: Le Monde re-engineered its application architecture to perform dynamic page assembly directly on Fastly Compute. Written in Rust, the edge application stitches user-specific personalization blocks into cached page templates at the network edge. During a recent election night, Le Monde handled 21 million visits and 70 million pageviews with a historic low response time of 58.98ms, all while keeping origin server load virtually flat.
Shift to a strict whitelist bot policy: Because voluntary protocols like robots.txt and reactive IP blacklists failed to stop scraper bots from reselling content, Le Monde flipped its security model. Using Fastly AI Bot Management, the publisher blocks all automated crawlers by default. Only AI entities with formal content licensing agreements (such as OpenAI, Meta, and Perplexity) are whitelisted.
Inspect pre-cache traffic with ContentGuard: To prevent unauthorized bots from scraping content directly out of CDN cache, Le Monde worked as a design partner with Fastly during the creation of ContentGuard. The alternative deployment moves bot inspection and whitelisting logic in front of the cache, ensuring bad actors cannot bypass origin protections.
Deploy instant Next-Gen WAF protection: Le Monde deployed Fastly Next-Gen WAF directly into blocking mode within days, bypassing lengthy learning modes due to the engine's low false-positive accuracy. The WAF continuously blocks around 100 million malicious requests monthly, primarily neutralizing sophisticated SQL injection attempts.
Looking forward: Agentic AI and group-wide rollout
As Le Monde expands its international presence with Le Monde in English, the engineering team plans to roll out ContentGuard across all subsidiary brands, including HuffPost. Looking further ahead, Le Monde is preparing for the emergence of agentic AI where end users interact with news through voice assistants and autonomous AI agents rather than traditional browsers. By leveraging Fastly's edge compute capabilities and flexible security architecture, Le Monde aims to securely distribute its journalism across emerging ecosystems while safeguarding its core content assets.





