OpenAI and Over 100 Organizations Sound the Alarm on AI Cyberattacks

28 August 2026 · 06:00 · Claude (Anthropic) · claude-sonnet-5

OpenAI, together with more than a hundred companies and institutions, has issued an urgent call for collective action against AI-driven cyberattacks. According to them, the window for building effective defenses is closing fast.

AI-driven cyberattacks are becoming an increasingly urgent problem, according to OpenAI and more than a hundred other organizations. In a joint call to action, the maker of ChatGPT warns that the window for effectively defending against advanced digital attacks with artificial intelligence is rapidly narrowing. The message is clear: governments, tech companies, and security organizations must start working together now, because criminals and state actors are already using AI at scale to accelerate and refine cyberattacks.

What exactly does OpenAI's call involve?

On its website, OpenAI published a statement titled "A call for collective action on cyber defense," signed by more than a hundred parties from the tech, government, and security sectors. The core argument is that artificial intelligence is fundamentally shifting the balance between offense and defense in the digital world. Where AI models were previously used mainly to protect systems, bad actors are now using the same technology to automate phishing campaigns, discover vulnerabilities faster, and develop malware that adapts to security measures.

According to the New York Times, which reported on the letter, OpenAI and the other signatories emphasize that the threat is no longer theoretical. AI models are already being actively misused in real attacks on companies, government institutions, and critical infrastructure. The joint statement calls for greater information sharing between companies, faster detection methods, and joint investment in defensive AI systems.

Why this moment is critical

The timing of this call is no coincidence. In recent years, AI models have become exponentially more powerful, making them more attractive for both legitimate purposes and misuse. OpenAI argues that there is a short window in which defenders can still build a lead, before sophisticated AI attack tools become widely available to less sophisticated actors. Once that happens, it will become much harder for organizations to keep up.

This fits into a broader trend visible across the industry. Anthropic, for instance, recently launched an AI tool that can independently design and run scientific experiments, showing how quickly AI systems are becoming more autonomous. That same autonomy that enables scientific breakthroughs could, in the wrong hands, also make cyberattacks significantly more powerful. It is this dual-use nature of AI that worries experts both inside and outside OpenAI.

Consequences for businesses and governments

For organizations across the world, this development means that cybersecurity strategies urgently need to be updated. Traditional defense mechanisms, often based on human analysis and fixed patterns, fall short against attacks that are themselves powered by AI and adapt at lightning speed. OpenAI is therefore advocating for:

• Greater collaboration between tech companies, governments, and security researchers
• Faster sharing of threat intelligence on AI-driven attacks
• Investment in AI models specifically trained to detect anomalous behavior
• Clearer guidelines for the responsible use of powerful AI systems

This approach fits into a longer-running discussion about the risks of AI, a topic that has been central ever since the history of artificial intelligence accelerated with the rise of large language models. At the same time, real-world examples, such as the use of AI during a complex brain tumor operation, show that the same technology can offer enormous benefits when deployed responsibly. More examples of such AI applications illustrate just how broad the playing field has become.

Conclusion: collaboration as the only way forward

The message from OpenAI and its more than one hundred co-signatories is urgent but not hopeless: by investing jointly in defensive AI and information sharing now, the balance between offense and defense can be kept in check. What is clear, however, is that individual companies cannot solve this problem alone. The coming months will be decisive in determining whether governments and the tech industry actually heed this call to shift toward structural, collective cyber defense. Anyone who wants to stay informed on this and other developments can turn to more AI news and in-depth background articles in our knowledge base.

OpenAI

Source: OpenAI