What AI Can, Cannot, and Should Not Do
Akamai Technologies, Monday, September 21st, 2026
Akamai CEO Tom Leighton argues AI needs human guardrails despite its benefits, citing internet security lessons.
In this post, Akamai CEO Tom Leighton argues that while AI can meaningfully benefit humanity through medical research, business efficiency, and communication, it lacks the moral judgment needed for fully autonomous decision-making.
Drawing parallels to lessons learned from internet security, he says AI must remain under human control. Akamai commits to strict guardrails on its own AI use, including sandboxing and limits on agent access, while also strengthening defenses against AI-enabled attacks.
The post calls for collaboration with industry partners and governments to build safeguards ensuring AI benefits rather than harms society.