AI is reshaping courtrooms, but not everyone is convinced the shift is a good thing. A new report finds court professionals save about three hours per week using AI tools, with that number expected to climb to 9.2 hours within five years. Yet 44% of respondents worry over-reliance on AI will lead to more mistakes and erode public trust. Only 13% of courts have developed an AI-literacy strategy for their employees, leaving a significant gap between adoption and preparedness.
Legal tech company Clio is pushing into the judiciary space, trademarking the phrase "judiciary-grade AI" and acquiring Learned Hand to bolster its court-focused products. A 50-state survey reveals a patchwork of policies: Massachusetts imposes the strictest ban on AI for legal research and writing, while New York permits approved tools under certain conditions. Across the board, every state agrees that judges cannot hand off their decisions to AI.
Meanwhile, cybersecurity threats are accelerating at an alarming pace. Nearly 37,000 security flaws were reported in the third quarter of 2026, tripling from a year earlier. A recent cyberattack on European supermarkets forced retailers to throw away millions of dollars in spoiled food. Binance announced plans to boost investment in AI-enhanced detection systems following a $6.3 billion cryptocurrency theft. The Port of Rotterdam is constructing an AI-powered cyber shield, and Siemens is testing advanced AI models for defensive security.
On the consumer side, insurance customers are warming up to AI for simple tasks but draw the line where errors matter. About 30% have used AI in the past six months for quoting, claims, or service. More than half of AI-driven claims interactions still end with a human handoff, showing that trust remains a major factor. In the hotel industry, the AHLA reports that AI helps staff focus on guest experiences rather than mundane work, boosting efficiency without replacing employees.
Saab has launched the Sferical AI supercomputer in Linköping, Sweden, built using Nvidia systems to train AI agents capable of piloting aircraft. The supercomputer is backed by Wallenberg family companies, and AI-controlled unmanned combat aircraft may fly alongside manned Gripens before 2030. In a different corner of the tech world, San Francisco startup Subvocal raised $1.3 million for AI glasses that read facial muscles to convert silent speech into text. The glasses are available for preorder at $499, with deliveries expected to begin in March 2027.
European activists are pushing back against the rapid growth of AI infrastructure. More than 100 campaign groups across 15 countries are organizing protests against datacenters, with actions planned in Warsaw, Dublin, Copenhagen, Amsterdam, and London. Carbon dioxide emissions from AI datacenters are projected to reach six times 2023 levels by 2030. In the UK alone, 315 datacenters are queued to connect to the power grid, representing 73 gigawatts of demand against the country's peak of 45GW.
On the policy front, incoming Democratic majorities in Washington plan to move past the hands-off approach to AI advancement. Voters elected candidates who campaigned on AI guardrails, with 56% saying AI and data centers were a key voting factor. Democratic leadership intends to hold public hearings and pass legislation with reasonable guardrails, following dozens of state and local ballot measures that already limit new data center construction.
Key Takeaways
- Court professionals save about three hours per week using AI, but 44% worry it will increase mistakes and erode trust.
- Clio has trademarked "judiciary-grade AI" and acquired Learned Hand, while a 50-state survey shows wide variation in judicial AI policies.
- Security flaws reported in Q3 2026 tripled year-over-year to nearly 37,000, and Binance plans to boost AI detection after a $6.3 billion crypto theft.
- About 30% of insurance customers have used AI in the past six months, but over half of AI claims interactions still require a human handoff.
- Saab launched the Sferical AI supercomputer, built with Nvidia systems, to train AI agents for piloting aircraft potentially by 2030.
- Subvocal raised $1.3 million for AI glasses that convert silent speech to text, available for preorder at $499 with March 2027 deliveries.
- Over 100 activist groups across 15 countries are protesting AI datacenters as emissions projections and power demand soar.
- Incoming Democratic leadership in Washington plans AI guardrail legislation after 56% of voters cited AI and data centers as key voting factors.
- Only 13% of courts have developed an AI-literacy strategy for employees despite growing adoption.
- The Port of Rotterdam and Siemens are both deploying AI-powered defensive security systems in response to escalating cyber threats.
AI saves time in courts but risks eroding professional skills
A new report shows AI is saving court workers about three hours per week, with expectations to reach 9.2 hours within five years. Half of court professionals say caseloads have grown while 74% report backlogs staying the same or growing. However, 44% of respondents worry AI will increase mistakes and hurt public trust. The top concern is over-reliance on AI weakening professional skills, cited by 34% of respondents. Only 13% of courts have developed an AI-literacy strategy for employees.
Clio pushes AI for judges as states show different policies
Legal technology company Clio is developing AI tools for the judiciary and has trademarked the phrase judiciary-grade AI. The company hired Pablo Arredondo and acquired Learned Hand to build on existing court AI products. A 50-state survey found that policies on judges using AI vary widely with no national consensus. Massachusetts has the strictest ban on AI for legal research and writing while New York allows approved AI tools under certain conditions. All states agree judges cannot delegate their judicial decisions to AI.
AI sharpens cyber security as companies boost defenses
Companies worldwide report that AI is accelerating and complicating cyber threats. Nearly 37,000 security flaws were reported in the third quarter of 2026, tripling from a year earlier. A recent cyberattack on European supermarkets forced retailers to discard millions of dollars in spoiled food. Binance said it will increase investment in AI-enhanced detection systems after a $6.3 billion cryptocurrency theft. The Port of Rotterdam is building an AI-powered cyber shield while Siemens tests advanced AI models for defensive security.
Insurance customers embrace AI for simple tasks but resist where errors matter
Nearly 30% of auto and property insurance customers have used AI in the past six months for research, quoting, service, and claims. Customers accept AI when they can verify or undo the outcome but resist it when they cannot. Policy changes rank lowest in adoption across every generation. About 42% of customers who used AI while quoting said it influenced their purchase. More than half of AI claims and service interactions still ended with a human handoff.
CISO Ryan Davis says the unknown is where security meets reality
Ryan Davis is CISO at New Charter Technologies after transitioning from a young hacker to a cybersecurity professional. He believes leadership is inherent to a person and that diplomacy is a CISO's most important trait. Davis stresses the value of diversity on security teams including different genders and backgrounds. His biggest security concern is the unknown because it is where security meets reality. He advises investing in oneself and never stopping learning.
Buffer Depth Key to Scaling AI Across Multiple Sites
Buffer depth is a critical factor in how fast AI jobs finish in multi-site clusters. Increasing buffer depth to 1.5x BDP can eliminate packet retransmissions. This reduces latency and improves performance for low-latency applications like real-time analytics and machine learning. Scaling AI beyond the data center unlocks new possibilities for businesses.
AI Pilots Trained on Supercomputer to Fly Fighter Jets
Saab has launched the Sferical AI supercomputer in Linköping, Sweden to train AI agents for piloting aircraft. The AI agents can undergo many more training cycles than human pilots and can take over flight control in certain situations. The supercomputer was built using Nvidia systems and is backed by Wallenberg family companies. AI-controlled unmanned combat aircraft may fly alongside manned Gripens before 2030.
Democratic Majorities Plan New AI Rules and Oversight
Incoming Democratic majorities in Washington plan to end the hands-off approach to AI advancement. Voters elected candidates who campaigned on AI guardrails, with 56% saying AI and data centers were a key voting factor. Democratic leadership plans public hearings and legislation with reasonable guardrails on AI. This follows dozens of state and local ballot measures limiting new data center construction.
AHLA Report Says AI Helps Hotel Staff Serve Guests Better
The American Hotel & Lodging Association and Actabl released a report showing AI benefits hotel workers. The report says AI helps staff focus on guest experiences instead of mundane tasks. It highlights increased efficiency, improved customer service, and enhanced productivity. The report emphasizes that AI supports hotel staff rather than replacing them.
European Activists Protest AI Datacenters in 15 Countries
More than 100 campaign groups across 15 countries are planning two days of protests against datacenters. Actions include demonstrations in Warsaw, Dublin, Copenhagen, Amsterdam, and London. By 2030, carbon dioxide emissions from AI datacenters will reach six times 2023 levels. In the UK, 315 datacenters are queued to connect to the power grid, representing 73 gigawatts of demand against the countrys peak of 45GW.
Subvocal AI glasses read facial muscles to turn silent speech into text
San Francisco startup Subvocal developed AI glasses that use sensors and artificial intelligence to detect tiny facial muscle movements when a person mouths words silently. The glasses offer an alternative to typing and voice dictation, especially in places like offices, libraries, and public spaces where speaking aloud is inconvenient. The technology was first built as a prototype at Stanford University's TreeHacks hackathon in February. The startup raised $1.3 million in funding, and the glasses are available for preorder at $499 with deliveries expected to begin in March 2027.
Sources
- AI in courts: Time saved, and the skills courts risk losing
- Clio Wants To Bring AI to the Judiciary, But 50-State Survey Reveals Policies Differ in Whether They Allow It
- AI Sharpens Cyber Security as Firms Boost Defensive Spending
- AI Insurance Experience Study Webinar Insights
- CISO Conversations: Ryan Davis – The Unknown Is Where Security Meets Reality
- Scaling AI Beyond the Data Center: Buffer Depth in Multi-Site AI
- A supercomputer is training AI agents. They will replace humans in fighter jets
- Incoming Democratic majorities put AI giants on notice
- AHLA Report: AI frees hotel staff to focus on guests
- Activists across Europe prepare two days of protests against datacentres
- AI glasses that can read your facial muscles: How Subvocal's technology turns unspoken words into text
Comments
Please log in to post a comment.