Akamai report reveals autonomous AI attacks use high-volume legacy techniques

The shift toward autonomous AI attacks forces a re-evaluation of security thresholds that were originally tuned to minimize noise from human-led intrusions. For streaming providers managing massive CDN footprints and cloud metadata endpoints, the primary risk is no longer a sophisticated 'zero-day' but a zero-fatigue attacker that can brute-force thousands of targets simultaneously. This trend suggests that defense-in-depth strategies must move away from behavioral detection, which can be buried by high-volume noise, toward strict reachability controls and hardened identity management. As these agents become more prevalent in offensive evaluations, organizations should watch for a shift in security vendor roadmaps toward automated, real-time IP allowlisting and credential isolation to counter non-human persistence.

Akamai has positioned itself as a key voice in CDN and edge security, and its research into autonomous AI agents reflects a broader industry reckoning with non-human threat actors. In early 2025, Akamai reported a 158% year-over-year increase in API attacks targeting media and entertainment companies, underscoring that streaming infrastructure is already a primary target for automated exploitation. The company's State of the Internet series has consistently flagged that bot-driven and scripted traffic now accounts for the majority of malicious requests against video delivery platforms, setting the stage for the more sophisticated autonomous media analytics behaviors documented in the Road Runner report, which aligns with broader scaling AI agents trends. Recent autonomous AI agents have further demonstrated the scale of these threats, prompting new AI AGENT Act regulation discussions, while OWASP GenAI security standards continue to evolve to address these risks. To mitigate these threats, providers are increasingly adopting AI bot traffic controls to secure their edge environments, while also addressing agentic AI subscriber data risks, a concern echoed by risks from autonomous AI agents.