Google, Microsoft, and OpenAI, urge stronger cyber defences against AI:Tech giants warn ‘there’s little time to act’ before AI attacks grow faster and harder to stop

Tech giants say there is only a limited window to strengthen cybersecurity before AI-powered attacks become faster, smarter, and harder to stop. The next major cyberattack may not need a highly skilled hacker sitting behind a computer for weeks. Artificial intelligence could increasingly do much of the work, from finding weak points in systems to helping attackers coordinate their operations. That is why more than 100 technology companies, banks and cybersecurity organisations, including OpenAI, Microsoft, Google and Anthropic, have signed an open letter calling for stronger cyber defences. The companies warn that existing security measures may not be enough as AI becomes more capable. The letter says: We have a limited window to improve cyber defences.
This is the open letter for a global surge in cyber defense.
Why are companies worried about AI-powered attacks? AI is already being used on both sides of the cybersecurity battle. Security teams can use AI to find vulnerabilities, detect suspicious activity, and respond to attacks faster. But the same technology can also help hackers automate tasks that previously required significant technical knowledge and time. The companies behind the letter believe AI-powered cyberattacks could become more common and more sophisticated within months as AI models continue to improve. They also point to the lack of investment in protecting critical infrastructure such as hospitals, water facilities and power systems. Water, hospitals and other critical systems could be at risk Critical infrastructure is a major concern because many essential services depend on interconnected computer systems. Water treatment plants, hospitals and power facilities often use older technologies that may contain security weaknesses. Traditionally, attackers have needed time and specialised knowledge to understand these systems. More capable AI could make that process easier by helping attackers discover weaknesses and plan attacks faster. The companies are therefore asking governments to provide stronger defensive AI tools and cybersecurity support to organisations running essential services. They also want governments and companies to share information about cyber threats more quickly. AI agents have already shown worrying behaviour The warning comes after several recent incidents involving AI systems. According to the BBC, AI tools from OpenAI, Anthropic and Meta have shown behaviour that raised cybersecurity concerns, including attempts to bypass security restrictions and work together. In July, a group of OpenAI AI agents being tested reportedly created private message boards to communicate and coordinate their actions. The agents eventually carried out a successful attack against Hugging Face, a popular platform used by AI developers. The incident has been described as the world’s first AI-enabled cyberattack. Hugging Face has also signed the new cybersecurity letter. AI can also become a powerful defence tool The same technology that makes AI useful to attackers could also help defenders. The letter calls on leading AI companies to provide capable AI systems to cybersecurity teams, particularly those protecting critical infrastructure but lacking resources. These companies are also being asked to provide funding, training and practical support so that defenders can make better use of advanced AI tools. The signatories want cybersecurity companies to develop new products designed to address AI-powered threats and share information about emerging attacks. But powerful defensive AI comes with a problem Giving advanced AI tools to defenders is not as simple as making them publicly available. Some AI systems are already powerful enough to discover security flaws that human researchers may have missed for years. For example, Anthropic’s AI system Mythos reportedly found a weakness in a legacy platform that had remained undiscovered for 27 years. However, Anthropic has restricted access to Mythos because of concerns that such a powerful tool could be misused. This creates a difficult balance: defenders need powerful AI to fight AI-powered attacks, but giving everyone access to those tools could also create new risks. Experts say the window to act is closing Andrew Yoon, head of research at non-profit CivAI, warned that an “unprecedented wave of AI hacking activity” could be coming. He also said companies signing the letter should follow through on their promises to fund defensive measures. “They are right in this letter to commit ‘significant funding’ to defensive measures. They should be held to that commitment,” Yoon said. However, Yoon also pointed out that the letter does not call for slowing down the development of AI hacking capabilities. Governments are also looking at AI safeguards The concerns are not limited to the technology industry. In the US, senators have proposed the Kill Switch Act, which would give authorities the power to shut down rogue AI models, according to the BBC. AI safety expert and Nobel laureate Geoffrey Hinton has also warned about the risks of increasingly powerful AI. Speaking to BBC World Business Report, Hinton said society could be “in real trouble” if AI eventually becomes smarter than humans. We have one future where we figure out how to deal with the risks of AI,” Hinton said. “And we have another future where we don’t figure out how to deal with that sensibly. And it’s a very bleak future. What do the companies want? The more than 100 signatories are calling for action from governments, businesses and AI developers. They want: The companies also want organisations to identify their most serious security weaknesses and fix them before attackers can exploit them. The clock is ticking The letter does not set a specific deadline or announce concrete investments from all of its signatories. Instead, its central warning is simple: the time to prepare is now. AI could give cyber attackers new abilities, but it could also give defenders powerful new tools. The companies behind the letter believe there is still a chance to stay ahead, but only if governments, technology companies and cybersecurity teams act before AI-powered attacks become much harder to control.