
Behind the curtain — AI 2027: the results
The big picture: The disruptions of the past two weeks — from a global Microsoft outage to the FAA’s grounding of air travel to a consumer banking glitch at Chase — were just a small taste of what experts expect.
- Bridgewater Associates, the world’s largest hedge fund, put it bluntly in an assessment this month: “(We) manage money based on a 1 in 3 chance that an economically destabilizing attack will occur within the next year.”
Horacio Rozanski, Chairman and CEO of Booz Allen, calls it the “fog” Before war” – a foretaste of greater chaos to come, when the invading agents will reach their full potential and can attack “at the speed of thought”.
- “Now is the time to take as many precautions as possible,” Rozanski said. “Recognize that it happens.”
Most of these dysfunctions appear to be the result of attempts by businesses and governments to secure their IT and infrastructure systems against future large-scale cyberattacks. This pattern will persist.
- A lot future these will be real cyberattacks, carried out by US rivals or small malicious actors, using increasingly powerful AI to spot and exploit security vulnerabilities in society and in our daily lives. Most AI leaders and governments expect at least small-scale strikes and fear larger strikes.
The best case scenario described to us: regular, random outages and small-scale scares like losing access to a bank or traveling on a plane for a day.
- The reason: Every government agency and business will step up to upgrade their software systems. Errors are therefore inevitable.
- These are assumptions based on private intelligence and an understanding of AI’s capabilities, both to cause and solve problems.
What top managers are most concerned about: With limited expertise in AI, an understaffed government and large corporations are trying to keep pace with technology that is improving daily.
- Even if the government were staffed with competent experts, which it certainly is not, the scale of vulnerabilities across the federal, state, and local governments, not to mention across businesses, would make reliable security impossible.
think about it like a mole with infinite moles.
- Until last month, the White House and Congress were largely focused on security regulations and data centers, not looming cyber risks. Today, even the White House’s Super Intelligence Force (SIF) is scrambling to understand the growing risks while evaluating near-daily reports of possible breaches.
- And President Trump just ousted the White House’s national cybersecurity director and named Kirsten Davies, who was the Pentagon’s chief information officer, to replace her. Davies has deeper subject matter expertise than his predecessor, former RNC official Sean Cairncross.
To move back: Only by understanding the state of AI and cybersecurity can governments and businesses mobilize to prevent possible shutdowns of critical banking, electricity and water systems. This is our chance to distill our conversations in a digestible way for Axios readers:
- The double-edged sword. The good news: American AI labs (Anthropic, OpenAI, Google, etc.) have built the most powerful AI on the planet, especially for creating software. The bad news: AI that is great at creating software is also great at finding – and exploiting – its vulnerabilities, often called “bugs” or “holes.”
- The new dangers of old technologies. Government and business primarily run on computer systems designed for a pre-AI world. They are full of bugs and holes. A vulnerability – exploited by a bad actor, or even triggered by someone attempting a good-faith repair – can shut down an entire business or all air travel, as we just saw.
- China had planned this moment. A Beijing-sponsored hacker group, Volt Typhoon, has been hiding in America’s critical infrastructure – water, energy, transportation and communications networks – for more than five years. A sister group, Salt Typhoon, broke into major US telephone carriers and recovered call recordings and wiretap data. The total number of compromised systems is unknown. The administration’s view is that China is not slowing down and the United States cannot cede any advantages.
- America is trying – and struggling – to catch up. Remember when Anthropic held back its Mythos AI model because of its breathtaking and frightening cyber capabilities? Since then, some global companies, including JPMorgan Chase, have used Mythos and other technologies to locate and repair their vulnerabilities. Our intelligence agencies did the same. But most This is not the case for the American government and businesses.
- Cyber defense is expensive. Many government agencies and businesses simply cannot afford to protect their systems. Larger companies can outbid others for best-in-class cyber protection. The best companies are widely booked. Frankly, many companies are reluctant to waste millions unless they are convinced that a threat is imminent.
- The coming swarm. AI is getting better and better. And the ability of agent swarms to act autonomously increases every week. Case in point: The FBI warned on November 13 about ARTEX, free and open source software that allows hackers to unleash AI agents to execute multiple phases of an attack at machine speed. Texas’ energy regulator has sounded the alarm to the state’s oil, gas and utility networks, saying they are a target. The EU has warned governments against ARTEX after a 10-day cyberattack last month.
- The government is divided and distracted. This is an all-hands-on-deck moment, and most governments are busy pointing fingers or looking the other way. Trump shows little expertise in AI. Few people in the White House have a deep understanding of the technology or are working to mobilize the government to prevent a hypothetical attack. Congress is no better. A random person at the bar understands this subject with the same depth as many legislators.
- Wait-and-see attitude. Oddly enough, although many cyber-intelligent officials across different sectors or levels of government fear the same thing, they rarely talk about it to people outside their own domain and party. There is no real mechanism or institution to bring them together. Most simply assume that no serious preventative measures will be taken until a full-blown mini-disaster occurs. Yet they all assume a mini-catastrophe is brewing.
Reasons not to panic: Everyone we spoke to expects attacks. Few people predict a real catastrophe. There is a big difference between a bank failing for a day and a region failing for a week.
- A truly society-shattering attack that shatters pipelines or puts a region in the dark still likely requires a nation-state. The actors who can do so, primarily China, pay the highest price if they do: an attack on the US network is an act of war, which calls for a response against the aggressor’s systems.
- The fragmentation of America most likely isolates major incidents. Thousands of utilities and no single switch for the network mean that even a successful one tends to remain regional.
So what could we do? The solution lies in the very first point of our distillation: we have the technology necessary to provoke and eliminate this cyber threat. The government, through leverage or mandate, could round up AI companies and force them to help strengthen vulnerable U.S. systems before focusing fully on simply building more powerful models.
- Top executives at leading AI companies tell us they would happily comply if their rivals did. In fact, these leaders feel a sense of post-election urgency to act before a public event causes citizens to hate AI even more than they already do.
Jay Clayton, chairman of the White House task force, is finalizing a voluntary security regime for large AI companies and is likely best positioned to bring AI CEOs together. He is well-versed in cyberspace and views AI largely through a national security lens.
- Other solutions will come from the private sector. Jamie Dimon, chairman and CEO of JPMorgan Chase, chairs an Alliance for Critical Infrastructure (CIA), a group of financial, energy, technology, telecommunications and airline giants aimed at “strengthening national resilience, reducing systemic risk and maintaining economic continuity.”
The bottom line: When we lobby all of these government, business and AI leaders, they agree that the chances of working together to prevent an attack are lower than the chances of an actual attack.
Gn bussni