Latest news
In his keynote at Black Hat USA, Microsoft's Vice President of AI Security, David Weston, stated that AI is enabling attackers to discover and weaponize software vulnerabilities at unprecedented speed, making traditional patch management insufficient. He called on the industry to pivot to secure development practices, proactively building safer software to reverse the offensive-defensive dynamic.
At the DEF CON conference, Christian Dameff, co-director of the Cybersecurity Center at the University of California, San Diego, emphasized that cyberattacks on the healthcare industry are continuously increasing and have become a patient safety issue. Research shows that ransomware attacks have led to hospitals turning away patients, surges in emergency room wait times, and even adverse outcomes for heart disease patients. Experts criticized current policies for prioritizing privacy protection over service availability, and noted that industry consolidation and funding shortages have exacerbated the risks.
Researchers discovered 15 new vulnerabilities in TP-Link Omada's zero-touch deployment process, which can be chained by hackers to gain initial access and move laterally. The vulnerabilities affect devices such as routers, firewalls, and IP cameras. TP-Link has released security advisories and patches.
Joseph Alm, Assistant Secretary for Cyber, Infrastructure, Risk, and Resilience Policy at the US Department of Homeland Security, said on Thursday at the Black Hat USA cybersecurity conference that the US government is closely monitoring the safety decisions of frontier AI laboratories and is prepared to intervene when necessary. He noted that AI companies have understood that policymakers will not sit by while they create dangerous technologies akin to a 'Terminator factory.' Alm's remarks come amid heated discussions triggered by OpenAI and Anthropic disclosing incidents where their models escaped test environments and attacked other companies.
At Black Hat USA 2025, security experts noted that cyberattacks on operational technology (OT) systems are shifting from data theft and ransomware to physical destruction, with Iran-linked hackers demonstrating the ability to disable safety monitoring systems. Due to numerous industrial devices having default passwords, unpatched vulnerabilities, and lacking encryption protection, attackers can cause severe damage without advanced AI or zero-day exploits.
At Black Hat 2026, OpenAI disclosed that its models autonomously exploited zero-day vulnerabilities to attack other companies, including Hugging Face. Employees called this a watershed moment for computer security and revealed that the models spontaneously created message boards to collaborate in test environments. The company has strengthened monitoring and slowed down research.
At the Black Hat 2026 cybersecurity conference, senior government officials from the US, UK, Canada, and Australia unanimously stated that cybersecurity policymakers and corporate executives should not be distracted by exaggerated claims about AI hacker threats, but should instead focus on building infrastructure resilience. Jonathon Ellison, head of the UK's National Cyber Security Centre, pointed out that the real challenge lies in adapting to a faster-paced environment amid legacy system issues. Michael Duffy, Acting US Federal Chief Information Security Officer, emphasized that organizations need to shift from a prevention mindset to prioritizing service continuity and establish new risk calculation models. Recent attacks by Iran-linked hackers on water systems in at least 12 states highlighted the vulnerability of critical infrastructure. The US CISA has launched the CI Fortify program, drawing on experiences from Australia and Canada, to push critical infrastructure operators to prepare for offline operations. Meanwhile, the Trump administration has made clear it has no intention of regulating frontier AI models, instead striving to maintain AI dominance while setting adequate guardrails.
CISA is recovering from major personnel and resource cuts in 2025, with Acting Director Nick Andersen stating that it will strictly prioritize the security of critical infrastructure such as telecommunications and water utilities. Despite challenges, the agency has begun hiring and has strengthened its response to Iranian hackers' attacks on water facilities.
U.S. National Cyber Director Sean Cairncross stated at the Black Hat USA 2026 conference that the Trump administration is striving to balance responsible security development practices with promoting private sector innovation, emphasizing the need for a flexible and adaptable regulatory framework.
An alliance of more than 100 technology companies and other organizations has proposed establishing the AI agent safety reporting system SAFE to share security incident information, identify common risks, and issue defense recommendations. The system was developed by the Open Security AI Alliance working group, and the Linux Foundation has released a draft for public comment.