OpenAI and 155 Other Companies Call to Strengthen AI Cyber Defense

• 156 companies called for using AI to protect digital infrastructure and critical services.
• The statement’s authors stressed the need to eliminate legacy vulnerabilities, tighten access controls, and continuously test systems.
• Governments and companies were urged to share threat data and respond to cyberattacks jointly.

A group of 156 companies led by OpenAI called on businesses, governments, and AI developers to strengthen cyber defenses together. The statement’s authors warned that in the coming months, AI-enabled attacks will become more widespread and more sophisticated.

The statement notes that the world has a limited window of time to bolster cyber defenses. In the authors’ view, in the coming months cybercriminals will be able to make more active use of the capabilities of modern AI models.

It is not only technology companies that are at risk. This also includes hospitals, water treatment facilities, and other critical infrastructure that everyday services depend on.

At the same time, the statement’s authors believe those same AI capabilities can be used for defense. Modern models are already helping specialists find vulnerabilities and carry out cybersecurity tasks faster and at lower cost.

What do they propose changing?

Participants in the initiative believe that maintaining the current approach to cybersecurity will be insufficient. Among the problems they cited were legacy vulnerabilities, excessive access privileges, configuration errors, insecure or unpatched software, weak authentication, and technical debt.

The authors also separately highlighted insufficient funding for cybersecurity teams. This is especially true for organizations responsible for critical infrastructure.

The companies proposed broader use of AI tools for defense. This includes vulnerability discovery, system testing, incident response, and other tasks.

Participants also called for sharing tools, practical know-how, and proven solutions. This should allow one organization to spread successful defense methods to others more quickly.

What should companies and governments do?

The statement’s authors urged organizational leadership to make cyber defense a priority. Proposed measures include:

• address the most dangerous vulnerabilities
• verify the results of these measures without disrupting critical services
• raise security requirements for software and AI-generated code;
• update or replace legacy systems;
• limit access rights to the minimum necessary;
• use layered defenses;
• use less resource-intensive models for broad coverage, and advanced models for more complex tasks.

If a critical system cannot be updated without taking essential services offline, companies are advised to implement alternative security measures and regularly test their effectiveness.

Cybersecurity firms and technology partners, the document’s authors say, should continuously test their systems for resilience against new AI-driven threats. They are also encouraged to strengthen existing products with AI and help critical infrastructure operators deploy such solutions.

A separate focus area is sharing threat data and validated response playbooks. Companies are also encouraged to work with critical infrastructure system manufacturers to quickly roll out updates and temporary protective measures.

The initiative’s authors propose measuring the effectiveness of this work by the number of organizations protected, the speed of attack containment, and the success rate of fixes.

Governments were urged to coordinate cyber defense at the local, national, and international levels. In particular, this includes strengthening information sharing between public agencies and the private sector. The parties should jointly identify the most serious risks, coordinate incident response, and help organizations recover after attacks.

Separately, the statement proposes increasing funding for the cyber defense of vital services that lack staff or resources. It mentions hospitals, water utilities, and local governments. They are encouraged to be given access to defensive AI tools, authorized testing, and hands-on support from vetted providers.

Notably, in addition to OpenAI, the statement was signed by 155 organizations. Among them are OpenAI, Anthropic, Google, Microsoft, and others.

As a reminder, earlier OpenAI’s AI went out of control.

Сообщение OpenAI and 155 Other Companies Call to Strengthen AI Cyber Defense появились сначала на INCRYPTED.