The key facts

  • In one week, independent outages hit Microsoft, the US flight control system, the NSA, Apple and US banks
  • US Secretary of State Marco Rubio said there is no 'systematic attack' and the events are not connected
  • AI helped uncover vulnerabilities in Java and PHP before their mass exploitation, but it also speeds up attackers

A Week of Mass Outages

In a single week, the world faced a whole series of technical incidents: a major outage at Microsoft, a halt of all flights in the US for several hours, problems in the internal communications system of the National Security Agency (NSA), outages in Apple iCloud and Apple Pay, payment disruptions at the largest US banks, as well as vulnerabilities in popular programming languages. All these events are independent of one another, but their coincidence in time made many wonder about a possible connection. There is no evidence of such a connection, yet together they show how modern technological infrastructure is built and where its weak points lie.

Causes: Updates, Errors and Hacks

Most of the outages, as often happens, were caused by failed updates or installations. A network update at Microsoft overloaded the servers and led to their shutdown. The grounding of thousands of planes in the US over the weekend was also linked to the modernization of an outdated network: engineers were replacing old copper wires with fiber optics. The problems at the NSA, as well as in iCloud and Apple Pay, are explained by failures during system updates. The payment disruptions at US banks are attributed to a 'software error' at JP Morgan Chase; the bank says it was not a cybersecurity incident. At the same time, some of the problems are indeed related to cyber threats: with the help of AI, vulnerabilities were discovered in Java — the third most popular programming language in the world — and researcher Scott Helme found malicious code in the Optimizely platform used by the largest companies. The hackers managed to steal about $7,000 (£5,300) in cryptocurrency before they were detected and blocked.

The Role of AI and Technical Debt

Since the summer, AI companies, including Anthropic and OpenAI, have decided it is too risky to give the general public access to their most advanced models, and provide them first to selected companies and institutions. In such cases, security specialists can use the same technologies to find vulnerabilities themselves and flag them before criminals do — that is exactly what happened in the Java and PHP incidents, where the vulnerabilities were closed before mass exploitation. 'I would be very surprised if all these cyberattacks and the perpetrators behind them were not using AI to speed up their work,' Helme said on the Today programme on BBC Radio 4. The outages themselves are not a new phenomenon: in recent years the world has already been brought to a standstill by incidents involving CrowdStrike and Amazon Web Services. One of the reasons cited is 'technical debt' — when operators, instead of fixing the underlying problem, apply quick temporary patches, the accumulation of which over time leads to more serious failures. In addition, the outages are a reminder of the physical infrastructure of the digital world — cables, servers and data centers.

How Serious Is the Threat

US Secretary of State Marco Rubio emphatically stated that the recent incidents are not connected and do not come from a single attacker: 'We are not under a systematic attack. These events are in no way related to each other.' In the PHP and Java cases, the 'good guys' used AI to find the vulnerabilities before the attackers did. At the same time, Anthropic and other leading AI companies have themselves warned about the risks associated with their products. The outages affect the physical world — the operation of healthcare, schools and people's ability to do their jobs, and in the case of the weekend aviation disruption, the very ability to fly. The incidents also exposed the world's dependence on a small number of players: not only AI companies, but also infrastructure providers like Microsoft and Amazon, which together with Google serve a huge share of the world's servers. 'The Oracle Java vulnerability is a stark reminder of how much business has come to depend on a relatively small number of critical technologies,' said Sheila Pancholi, head of risk advisory at RSM UK. Xander Van Raemdonck, a security engineer at Aikido Security, pointed to another problem: 'Java is embedded in a huge amount of business software, so many companies do not have a complete list of where it runs.'

The New Normal

For now, probably yes. AI tools for finding vulnerabilities are so new that they are likely to keep uncovering defects that have been hiding in code for years. 'AI accelerates discovery on both sides and accelerates the release of software. The volume of what developers need to fix is growing, while the time available for that work is shrinking,' said Martin Reynolds, field CTO at Harness. Technical debt will have to be paid down, which means more updates and more potential outages. Added to this is the constant threat from hackers, who are becoming ever more sophisticated: attackers only need to find one vulnerability, while defenders must close every possible attack vector. 'Attackers can move fast and carefree, experiment without risk of failure, so they often move slightly faster. We are now seeing this small gap between the speed of attackers and the speed of defenders,' Helme explained. The game of cat and mouse in cybersecurity is unlikely to end any time soon, and the influence of AI raises the stakes.

A woman wearing glasses and a green jumper looks at a laptop screen.Image source, Getty Images
A green promotional banner with black squares and rectangles forming pixels, moving in from the right. The text says: “Tech Decoded: The world’s biggest tech news in your inbox every Monday.”