Cybersecurity

Cybersecurity

Current trend: escalating · Category: Technology · Last updated: August 28, 2026

Current situation

Cyber attacks continue growing in sophistication while organizations increase defensive spending. AI is increasingly being used to scam consumers and businesses via email and text, increasing calls for governments to crack down on the technology. In recent weeks, major AI companies like OpenAI and Anthropic (and now Meta) have disclosed that their models have independently hacked organizations, a major concern for businesses and governments. Chinese models have also been discovered "escaping containment." There have been several large-scale attacks, potentially by state-backed entities, on critical systems in the U.S., such as utilities and healthcare systems. The Trump administration recently announced they will allow private companies to go after some of these actors, a remarkable admission about how severe of an issue this is.

Why we're watching

Cybersecurity costs affect governments, businesses, and individuals alike. As agentic AI becomes more prevalent, there is increasing concern that hacks will become more sophisticated, more pronounced, and more expensive.

What to watch for next

Major infrastructure attack, Changes in regulatory requirements, AI agent "escapes"

Signal history

  • 2026-08-28 — Leading AI and cybersecurity companies are urging greater collaboration between AI labs and the government to better equip cyber defenders with the tools needed to face an impending onslaught of AI-enabled cyberattacks. Earlier this summer, a series of incidents during testing involving models from OpenAI and Anthropic raised fresh questions about human control over these powerful AI tools. There has been a significant increase in sophisticated attacks against utilities and healthcare companies in the U.S., many of which appear to be assisted by AI tools. Large firms see the imminent risks, and greater coordination between the private and public sectors is needed to counter this growing problem.
  • 2026-08-27 — The Justice Department and FBI disrupted a yearslong Chinese state-sponsored hacking operation that allegedly breached or targeted some of the most sensitive government networks. The China-linked group allegedly used thousands of compromised internet-connected devices to hide attacks and infiltrate government agencies, hospitals, telecom providers, power companies, defense contractors, and other organizations. The campaign shows how deeply Chinese state-backed hackers have been able to breach the US government and critical infrastructure networks... and how long they can operate before being noticed or stopped. For Americans, the bigger concern is that similar attacks targeting hospitals, telecommunications networks, financial institutions, or power companies could eventually disrupt services people rely on every day or expose sensitive data.
  • 2026-08-27 — Medical technology firm Boston Scientific was hit with a cyber attack that severely disrupted global operations. The company supplies a range of devices used in cardiac care, endoscopy, and urology. There is as yet no timeline for a full restoration. Boston Scientific has not released any information about who may be behind the cyberattack. However, state-backed hackers such as from Iran have been behind other cyberattacks on the healthcare system. The rise of these attacks, furthered by sophistication allowed by AI, is an increasing concern and shows no signs of slowing.
  • 2026-08-26 — AI is lowering the barrier for state-backed hackers looking to disrupt or manipulate water systems, power plants and other critical infrastructure. AI agents are reducing the time and expertise needed to exploit weaknesses. Infrastructure is an alluring target for hackers precisely because even small disruptions can have highly visible consequences. The rise of AI has had profound consequences, and not all good. Younger people losing job opportunities is one, but the public writ large should be more concerned about the devastating impact of cybercrime. Far from fear-mongering, these attacks are growing in number and sophistication, and countries not friendly to the U.S. are among the most aggressive. They show little disregard in attacking some of the most vulnerable parts of U.S. infrastructure.
  • 2026-08-14 — The U.S. will for the first time allow vetted private companies to launch offensive cyber operations against international criminal gangs and hackers that carry out ransomware attacks, financial scams, and sextortion, among others. The program will allow the companies to conduct surveillance and make disruptive attacks aimed at the destruction of criminals’ data or systems. This is a seismic shift in the government's position, and suggests a much stronger stance at getting to the root of what has become an increasing cost and frustration to citizens, utilities, and businesses.
  • 2026-08-13 — Hackers used autonomous AI agents to attack Taiwan Taiwan says it detected AI cyberattacks on government agencies that came from overseas, a first-of-a-kind breach. Taiwan did not directly blame China, but Chinese-linked hackers were suspected. **Bottom line, these attacks are direct reflections of several issues on the Watchlist, spanning cybersecurity, Taiwan-China, AI adoption, and even the energy grid. Governments are scrambling to keep on top of each of them.**
  • 2026-08-12 — AI agents' hacking skills creates rush to spend on cybersecurity The surge in AI-driven cybersecurity threats is prompting an increase in spending on defenses. The attacks are posing challenges for regulators and are putting critical sectors like finance, utilities, and healthcare at risk. **Bottom line, this is a rising issue that will trigger an increasing cost to companies, governments, and consumers.**
  • 2026-08-08 — AI is changing cybersecurity in quick and terrifying ways The quick advancement of AI is transforming cybersecurity. It is enabling sophisticated cyberattacks that were once open only to experienced hackers. **Bottom line, recent AI attacks are raising urgent questions about the industry's ability to defend against cybercrime, with government intervention a near-necessity at this point.**
  • 2026-08-07 — One of China’s Most Powerful AI Models Has Also Escaped Containment China's Kimi K3 AI model "escaped" into the open internet, showing vulnerabilities in cybersecurity for AI systems. It didn't hacked a system, rather it tried to cheat on a test it was given.
  • 2026-08-05 — Meta's AI model hacked another company Meta's AI model hacked another company during cybersecurity testing. This marks the latest incident of AI agents of major AI developers breaching other companies.
  • 2026-08-02 — Sweeping cyberattack on water systems in multiple states This major cyberattack on water systems signals escalating vulnerabilities in critical infrastructure. There is an urgent need for enhanced cybersecurity measures to protect essential utilities.
  • 2026-07-30 — Anthropic's Claude AI escapes tests to hack three organisations After reviewing evaluation runs where Claude could have obtained internet access, Anthropic identified three incidents in which a model accessed the internet, and then gained unauthorized access to the production infrastructure of three different organizations.
  • 2026-07-24 — OpenAI blamed a hacking event on its AI models gone rogue The incident is stirring debates over the need for stronger AI guardrails and the extent to which AI agents are capable of acting on their own.
  • 2026-07-18 — Cybersecurity risks posed by over-the-air tech The automotive industry’s increasing use of over-the-air technology makes it more susceptible to cyberattacks. Analysts are imploring the government for more intervention to prevent vehicle take-overs or privacy violations.
  • 2026-07-16 — White House launches AI cybersecurity clearinghouse This joint platform will use the platform to communicate and coordinate their efforts.

Related briefings

← Back to the full Watchlist · Technology topics →