More than 100 companies call for a defensive surge against AI-driven cyberattacks

More than 100 technology and financial companies have called for a broad effort to strengthen cyber defences before more capable artificial-intelligence systems make digital attacks easier to automate and scale, according to Reuters.

The group includes OpenAI, Microsoft, Alphabet, Amazon, IBM and major financial institutions. Their concern is not that AI suddenly invents cybercrime. Attackers already scan networks, steal credentials, exploit software flaws and run phishing campaigns at enormous scale.

The worry is that AI lowers the cost of doing all of those things.

A less-skilled attacker can use a model to write convincing phishing messages, translate scams into multiple languages, analyse leaked code, generate scripts and search for weaknesses more quickly. A more sophisticated attacker can use AI to speed up reconnaissance and automate repetitive parts of an intrusion.

That creates an asymmetry. Attackers only need one useful weakness. Defenders have to protect thousands of systems, employees, suppliers and software dependencies at the same time.

Large organisations often make that problem worse through complexity. Security teams may operate across fragmented tools, legacy software and slow approval processes. An attacker can move quickly while a defender waits for a patch, budget decision or internal sign-off.

The coalition is pushing for stronger access to advanced models for vetted defenders and more leadership attention to cybersecurity. The idea is that the same capabilities that help attackers can also help defenders analyse alerts, review code, identify suspicious behaviour and respond faster.

But AI is not a substitute for basic security. Many successful attacks still depend on old problems: weak passwords, unpatched software, excessive permissions, exposed credentials and poor network segmentation.

A company that layers AI on top of weak infrastructure may simply detect problems faster without fixing the conditions that created them.

There is also a governance issue. Giving security teams access to more powerful models means organisations need controls around sensitive data, logging and model use. A cyber-defence system can itself become a high-value target if it has deep access to corporate networks.

The larger point is that the speed of offence is changing. AI can compress tasks that once took hours into minutes and allow attackers to run more experiments at lower cost.

Defence therefore has to become more automated too, but without losing human oversight.

The companies signing the call are effectively warning that the next cybersecurity gap may not come from a single breakthrough exploit. It may come from thousands of ordinary attacks becoming faster, cheaper and more convincing at the same time.