AI security concerns are sharpening as governments and infrastructure operators weigh faster-moving cyber threats. The Operational Technology Cybersecurity Coalition’s report offers 18 recommendations based on four summer roundtables with nearly 70 participants, including basic protections, plans for breached systems, and closer coordination between AI developers and critical-infrastructure operators. Legacy operational technology is a particular challenge: about 70% of OT vulnerabilities are in systems that cannot be patched or rebooted without disrupting operations.
Sens. Maria Cantwell and Mike Rounds want the White House’s Super Intelligence Force to lead a national response. They recommend collaboration among AI developers, security firms, infrastructure operators, NIST, CISA and Department of Energy national laboratories to identify vulnerabilities and develop fixes. They also call for risk-based standards for defensive AI and rules ensuring autonomous agents’ actions can be explained, tested and audited. Germany’s BSI reports its cybersecurity threat index rose 83% in 2026, from 130 to 238 points. Attacks exploiting previously unknown vulnerabilities increased 48%, and the average interval between public disclosure and exploitation fell from two months to eight days.
Companies are also working through the operational risks of upgrades. Newport cybersecurity company Nexigen CEO Jon Salisbury said advanced code-analysis and security tools from groups including Anthropic and OpenAI became available in April, prompting organizations to strengthen systems at the same time. That may help explain recent disruptions involving Microsoft Outlook, airport operations and Chase Bank, though experts say none has been linked to a cyberattack. In a separate discussion of AI safeguards, CTC CEO Chris Hamilton says controls should emphasize risk, independent testing and incident reporting. CTC bars staff from entering proprietary or controlled unclassified information into public or third-party generative AI models, and uses AI to analyze Department of War mishap data and test autonomous systems. Hamilton previously led Georgetown’s work on a $100 million follow-on effort for the Office of the Director of National Intelligence.
AI adoption is also changing workforce preparation and enterprise training. At Insights2040, banking discussions focused on building skills that complement automation, including creativity, empathy and complex problem-solving, through upskilling and reskilling and closer cooperation among employers, educators and governments. Another article describes combining AI, simulations and digital adoption tools to help employees practice software and complex business processes in virtual settings, with potential gains in engagement, productivity and customer satisfaction.
Evidence of AI’s real-world effects varies by application. A review of Montgomery County, Maryland, school-bus camera enforcement found little evidence that AI-powered cameras reduced crashes near stopped buses: citations per active camera averaged 36 in 2020 and 34 in 2025. Police reported 40,854 citations or warnings from Jan. 1, 2025, through Sept. 30, 2026. BusPatrol flags possible violations for police, who decide whether a violation occurred; the county says the program supports accountability, education and safety data collection. Separately, Utah prescription pilots that use AI to renew or write some prescriptions remain an oversight question after an HHS clinical AI roadmap did not clarify whether the FDA oversees them or which agency checks the systems.
Universities and public agencies are pursuing AI skills and applications. A University of Wisconsin–Milwaukee team, Machine Minds, took third place in a month-long physical AI hackathon with a recycling robot that identified and sorted objects and adapted when bins moved. The 270-student competition involved 42 teams and was hosted by UWM’s Connected Systems Institute and Milwaukee School of Engineering with Rockwell Automation and Nvidia; the winners received $1,000, an Nvidia RTX 5070 and $1,000 in Brev credits. Saudi Arabia’s SDAIA has launched its Year of Artificial Intelligence initiative at universities, with training, seminars and consultations intended to develop skills, promote responsible use and turn academic projects into practical solutions. Its first phase is expected to finish by year-end.
Police leaders will discuss AI’s potential role in policing at a sponsored webinar on Dec. 3, covering systems that connect information, support investigations and perform multistep tasks, as well as their limitations and readiness for daily use. Separately, the BSI warns that autonomous AI agents could enable larger-scale attacks, while the U.S. senators call for their actions to be explainable, testable and auditable.
Key Takeaways
- OTCC issued 18 recommendations for AI-era infrastructure security; about 70% of OT vulnerabilities affect systems that cannot be patched or rebooted without disrupting operations.
- Sens. Maria Cantwell and Mike Rounds urged the White House’s Super Intelligence Force to coordinate a national response and called for risk-based defensive-AI standards and auditable autonomous agents.
- Germany’s BSI says its cyber threat index rose 83% in 2026, while exploitation after public vulnerability disclosure now takes an average of eight days, down from two months.
- Nexigen’s Jon Salisbury linked clustered outages affecting Microsoft Outlook, airports and Chase Bank to simultaneous system upgrades after tools from Anthropic and OpenAI became available in April; experts have not linked the outages to cyberattacks.
- CTC CEO Chris Hamilton supports risk-focused AI safeguards, independent testing and incident reporting. CTC bars staff from putting proprietary or controlled unclassified information into public or third-party generative AI models.
- Hamilton previously led Georgetown’s work on a $100 million follow-on effort for the Office of the Director of National Intelligence.
- Banking and enterprise-training discussions emphasize reskilling, human strengths such as empathy and problem-solving, and simulations for practicing software and business processes.
- A review found little evidence that Montgomery County’s AI-powered school-bus cameras reduced crashes; citations per active camera averaged 36 in 2020 and 34 in 2025.
- HHS’s clinical AI roadmap leaves unclear whether the FDA or another agency oversees Utah pilots that use AI to renew or write prescriptions.
- UWM’s Machine Minds won third place in a 42-team physical AI hackathon with an adaptive recycling robot; Saudi Arabia’s SDAIA launched a university AI initiative, and police leaders have a Dec. 3 webinar on AI in policing.
OTCC calls for stronger resilience against AI cyberattacks
The Operational Technology Cybersecurity Coalition (OTCC) urged U.S. policymakers to make critical infrastructure resilience a key part of national AI security policy. Its report, “The Missing Pillar of AI Security: Critical Infrastructure Resilience,” offers 18 recommendations based on four summer roundtables with nearly 70 participants from industry, government, and cybersecurity groups. The report calls for basic security measures, planning for systems to be breached, and cooperation between AI developers and operators of critical infrastructure. It also highlights the challenge of protecting legacy systems, noting that about 70% of OT vulnerabilities are in systems that cannot be patched or rebooted without disrupting operations.
Senators urge White House to act on AI cyber threats
Sens. Maria Cantwell and Mike Rounds urged the White House’s Super Intelligence Force to lead a national effort against AI-enabled cyber threats. They said AI can find software weaknesses and help attackers exploit them faster than many organizations can respond. The senators recommended that AI developers, cybersecurity companies, critical infrastructure operators, NIST, CISA, and Department of Energy national laboratories work together to find vulnerabilities and develop fixes. They also called for risk-based guidance and standards for defensive AI tools, along with rules to make autonomous AI agents’ actions explainable, testable, and auditable.
Germany reports an 83% rise in cyber threat index
Germany’s Federal Office for Information Security (BSI) said the country’s cybersecurity threat index rose 83% in 2026, from 130 to 238 points. Attacks exploiting previously unknown software vulnerabilities increased 48%, while the average time from public disclosure of a vulnerability to its exploitation fell from two months to eight days. The BSI warned that autonomous AI agents may carry out more large-scale attacks, and said some vulnerabilities are exploited within hours of discovery. Interior Minister Alexander Dobrindt described the threat level as high and the outlook as critical.
Banks rethink how staff gain skills as AI changes work
Discussions at Insights2040 examined how banks can prepare workers for roles changed by AI. As AI takes on routine tasks, the traditional approach of learning on the job and gradually taking on more responsibility may no longer be enough. Employees need skills that complement AI, including creativity, empathy, and complex problem-solving. Banks can respond by investing in upskilling and reskilling, while employers, educators, and governments work more closely together.
AI and simulation reshape enterprise software training
Companies can combine AI, simulation, and digital adoption tools to improve enterprise software training. AI can automate tasks, support data-based decisions, and help create engaging learning experiences for employees. Simulations let employees practice software use and complex business processes in virtual settings that mirror real situations. The article says these approaches can improve employee engagement and productivity, as well as customer satisfaction.
Montgomery County bus cameras issue tickets, but safety gains remain unclear
A Bloomberg Businessweek and Type Investigations review found little evidence that AI powered school bus cameras in Montgomery County have reduced crashes near stopped buses. Citations per active camera changed only slightly, from an average of 36 in 2020 to 34 in 2025. The investigation identified at least three children struck while getting on or off stopped buses since 2017, based on local news reports and camera footage. Police reported 40,854 citations or warnings from January 1, 2025, to September 30, 2026, while Fairfax County recorded more than 6,300 from April 2024 to December 17, 2025. BusPatrol cameras identify possible violations and send video and images to police, who decide whether a violation occurred; Montgomery County’s Vision Zero program says the cameras support accountability, education and safety data collection.
UWM team earns third place with AI recycling robot at hackathon
University of Wisconsin Milwaukee team Machine Minds placed third in the month long Physical AI Hackathon, earning $1,000, an Nvidia RTX 5070 graphics card and $1,000 in Brev credits. The competition was hosted by UWM’s Connected Systems Institute and the Milwaukee School of Engineering with Rockwell Automation and Nvidia, and included 270 students on 42 teams. Machine Minds built a robotic system to identify objects, sort them into bins and adjust when the bins moved. The team tested its solution with a digital twin before using an industrial robot, which sorted all target objects in a demonstration and kept working after bin locations changed. UWM team VYFLARE also reached the finals and received $500 and $1,000 in Brev credits.
Experts link recent technology outages to AI driven system upgrades
Recent disruptions to Microsoft Outlook, airport operations and Chase Bank systems may be linked to companies upgrading systems after AI tools revealed flaws and weaknesses. Jon Salisbury, CEO and cofounder of Newport cybersecurity company Nexigen, said advanced code analysis and cybersecurity tools became available in April from groups including Anthropic and OpenAI. He said organizations are racing to strengthen their systems at the same time, which may explain why outages have clustered. Salisbury compared the upgrades to repairing potholes, saying the work is meant to improve systems but can cause temporary problems. Experts said none of the outages has been linked to cyberattacks, though some disruptions may be connected to efforts to prevent them.
Federal clinical AI roadmap leaves questions about Utah prescription pilots
The Department of Health and Human Services released a roadmap to speed the use of artificial intelligence in clinical care, but it does not explain whether the Food and Drug Administration oversees Utah pilot programs that let AI renew prescriptions and, in one case, write them. Utah residents can obtain some prescriptions through these systems for a few dollars. The article says there is no clear federal answer about which agency checks the systems before and after they launch. It frames that uncertainty as an accountability concern for patients. The available article text says the roadmap centers on state sandboxes but does not provide further details.
Police leaders to explore AI developments in December 3 webinar
A webinar on December 3 will examine how artificial intelligence may shape policing in 2027 and beyond. The event is scheduled for 11 a.m. ET, 10 a.m. CT and 8 a.m. PT, and organizers will send a recording to people who register but cannot attend. The discussion will cover AI systems that connect information from multiple sources, support investigations and carry out multi step tasks, as well as the limits of these tools and how close they are to daily use. The panel includes former police chief Ken Wallentine, University of South Carolina professor Ian T. Adams and Redmond Police Chief Darrell Lowe. The webinar is sponsored, and the article is dated November 19, 2026.
Saudi Arabia launches AI initiative across its universities
The Saudi Data and AI Authority (SDAIA) has launched its Year of Artificial Intelligence initiative at universities across the Kingdom. It includes seminars, panel discussions, training workshops, interactive booths and specialized consultations. The program aims to spread knowledge about AI, build national skills, encourage responsible use and turn academic projects into real-world solutions. SDAIA says the initiative supports Vision 2030 and the growth of the national AI ecosystem, with its first phase expected to finish by the end of the year.
CTC CEO Chris Hamilton discusses AI rules and security
Chris Hamilton, a former U.S. Army intelligence officer, became CEO of CTC after leading Georgetown University’s work on a $100 million follow-on effort for the Office of the Director of National Intelligence and spending 15 years at CMX Technologies, Xator and other organizations. He says CTC’s mission-focused culture, including its strong ties to veterans, attracted him to the organization. Hamilton argues that AI needs safeguards to protect sensitive information, but says rules are not yet fully in place and should focus on risk, independent testing and incident reporting. CTC prohibits staff from entering proprietary information or controlled unclassified information into public or third-party generative AI models. The organization uses AI to analyze Department of War mishap data and works to test the security and reliability of autonomous systems.
Sources
- OTCC urges US policymakers to bolster OT cybersecurity, fund resilience, govern AI agents as threats accelerate
- Cybersecurity Policy Report, AI’s Disruption of Cyber Defenses Demands White House Response, Senators Say, (Nov 18, 2026)
- AI-powered cyberattacks in Germany surge 83% as security threats escalate
- How are banks building expertise when AI changes the training ground?
- AI, Simulation and Digital Adoption: The Future of Enterprise Software Training
- Do Montgomery County AI-powered school bus cameras make children safer?
- UWM students showcase real-world AI skills in hackathon competition
- Tech experts say recent outages could be tied to artificial intelligence updates
- Federal Clinical AI Roadmap Leaves Unanswered Whether FDA Oversees Utah Pilots That Let AI Refill and Write Prescriptions
- Webinar: The next chapter of AI in policing: What leaders should be watching in 2027
- SDAIA Launches Year of Artificial Intelligence Initiative Across Universities in Saudi Arabia
- CTC CEO Chris Hamilton on AI Governance and Space Security
Comments
Please log in to post a comment.