- Applying lessons from the early internet
- Putting guardrails on AI use
- Our collaborative commitment to safety
AI is capable of doing things that can benefit humanity in many ways: accelerating medical research and the discovery of treatments that will extend lives, empowering businesses and agencies to function more efficiently and productively, and facilitating quicker communication across language barriers around the world. Even at the mundane level, AI agents offer the potential to give us time back for better things by handling everyday tasks on our behalf.
What AI agents cannot do better than humans is to make moral judgments about what is right and wrong. For that reason, AI must always be kept under human control.
As the CEO of an AI infrastructure company that the world’s largest enterprises and AI companies rely on, I want to add my voice to those who wisely call for proper checks and balances on the use of AI to better manage potential risks.
Applying lessons from the early internet
Managing the risks and unintended consequences of new innovations is not a new challenge for humanity or even for our company. A great example is what transpired with the invention of the internet.
As with the promise of AI today, the invention of the internet delivered massive benefits to humanity, including easy access to information, improved communication, and efficiencies that couldn’t have been imagined before it became a central aspect of all our lives. And as could happen with AI today, there are those who used the new technology to cause harm through fraud, piracy, the spread of disinformation, and the disablement of critical infrastructure and services.
Since 1998, Akamai has been at the forefront of making the internet fast, reliable, and secure for billions of people around the world, trillions of times a day. Our team of thousands of leading technologists has worked tirelessly to develop security measures to help keep the internet and its users safe from attacks of all kinds. And we go the extra mile by restricting customer use of our infrastructure to applications that do not cause harm and that are compliant with the laws of the countries in which we operate.
Putting guardrails on AI use
We intend to take a similar approach with AI, by requiring customers that use our platform for AI to follow the same rules of acceptable use that have long governed our network and that have helped to build Akamai’s hard-earned reputation for trust and integrity in everything we do.
Akamai is also actively developing sandbox protection for AI agents by providing guardrails on which systems AI agents can access. Limiting agents’ access to other systems on a need-to-know basis is just one of the many methods that purpose-built solutions can use to contain agents and prevent them from getting out of control, so they perform only what they are asked to do, under strict limits and watchful human control.
But putting guardrails on the use of AI is only part of the equation when it comes to making the world safe as AI proliferates. Our industry must devote just as much attention to protecting potential targets that agents could attack.
Defending systems against the misuse of AI requires having the strongest application firewalls, protecting your APIs, and having the ability to segment systems to contain breaches and stop deeper penetration and the exfiltration of data once defenses have been breached. And Akamai is working hard on the development of these cybersecurity capabilities to harden IT systems against attackers armed with AI.
Our collaborative commitment to safety
Throughout our company’s existence, Akamai has worked with public and private agencies and standards bodies to secure the internet and strengthen cybersecurity, to protect life online for all. We commit ourselves and our company to do the same when it comes to the use of AI and agents.
As Akamai has done successfully since our founding, we are also committed to working with our partners in industry, the public sector, and with the governments of the countries in which we operate to build and employ safeguards so that the use of AI will benefit humanity, not harm it.
