AI-generated political advertising is surging ahead of the election cycle. The Wesleyan Media Project tracked 220 AI-related midterm ads and over $137 million in spending. OpenAI is setting up an Election Operations Center to combat misinformation, while 31 states have passed deepfake laws. Experts advise voters to verify sources and approach emotional content with caution.
Some voters are turning to AI chatbots to learn about candidates, but experts warn these tools often produce misleading information. Tulane professor Aron Culotta recommends asking chatbots to show their sources. Voters generally report low trust in AI for political facts, and the accuracy problems seen during the 2024 election persist.
In the business world, AI agents are placing phone calls to local shops and restaurants in San Francisco. A test of 848 calls found about one in three people continued talking after learning they were speaking with AI. Some business owners find the calls annoying, while others are still evaluating whether this is a useful tool or a nuisance.
Cadence Design Systems launched Maestro, a suite of AI agents built for chip design. The platform uses specialized models and reinforcement learning to improve engineering workflows, working with ChipStack, InnoStack, ViraStack, and AuraStack. Cadence reports up to 30% better results and five times more token efficiency. NVIDIA is collaborating on agent security using Nemotron model technology.
The first fully AI-generated movie to receive a wide theatrical release flopped at the Chinese box office. "Sanxingdui: Future Past" earned roughly $198,000 despite over 10,000 opening-day screenings. The film cost about $1.5 million to produce. Audience members said the quality still falls short of human-made movies, though the production company called it an important test.
ISC2 is introducing a new AI security certification to meet growing industry demand. More than 7,000 volunteers are helping design the exam, with beta tests planned for early 2027. The certification aims to reach recognition similar to the long-standing CISSP. Experts also raise concerns about cyber warfare, AI risks, and quantum computing threats.
New research presents methods called CROG and CLIG to improve federated learning, where devices train shared AI models without sharing raw data. The approach weights each device's update based on how far it strays from a reference point, requiring no extra data, computation, or communication.
A Cyberfuturists report finds a gap between AI pilot projects and real production use in security operations centers. Five barriers stand in the way: trust, governance, integration, cost, and verification. Security teams often let AI summarize alerts but hesitate to let it take action.
Dr. Samuel Faulkner will speak on Feb. 8 about AI use in mental health, focusing on crisis care and ethical questions. He says AI is moving faster than research can verify whether it works safely. Meanwhile, University of Arkansas professors held an event in Fayetteville discussing both the benefits and risks of AI.
TÜV Rheinland spoke at the HFE & AI 2026 conference in Beijing on Oct. 17, 2026, advocating for human factors research to start at the beginning of product design. Expert Mia Xie argued that better technical specifications do not always lead to a better user experience.
Key Takeaways
- AI political ads have surged to 220 midterm spots with over $137 million in spending, according to the Wesleyan Media Project
- OpenAI is creating an Election Operations Center to fight misinformation ahead of the election
- About one in three people keep talking to AI agents after learning they are not human, based on 848 test calls in San Francisco
- Cadence launched Maestro AI agents for chip design, reporting up to 30% better results and five times more token efficiency
- NVIDIA is collaborating with Cadence on agent security using Nemotron model technology
- The first wide-release AI-generated movie, "Sanxingdui: Future Past," earned only about $198,000 despite costing $1.5 million to make
- ISC2 is launching a new AI security certification with beta tests planned for early 2027
- New federated learning methods CROG and CLIG improve AI model training across devices without extra data or computation
- Security teams face five barriers to production AI use: trust, governance, integration, cost, and verification
- Dr. Samuel Faulkner warns AI in mental health is moving faster than research can verify its safety
AI political ads surge as voters learn to spot fakes
AI-generated political content is growing fast ahead of the election. The Wesleyan Media Project found 220 AI-related midterm ads and over $137 million in spending. Experts say voters should check sources and be careful with emotional content. OpenAI is creating an Election Operations Center to fight misinformation. Thirty-one states have passed deepfake laws, but there are no federal rules.
Some voters use AI chatbots to decide elections
Some voters are using AI chatbots to learn about candidates and make election decisions. Experts say AI can save time but often gives wrong or misleading information. Tulane professor Aron Culotta says users should ask AI to show its sources. Voters say they do not fully trust AI for political facts. AI chatbots still face the same accuracy problems seen in 2024.
ISC2 launches new AI security certification
ISC2 is introducing a new certification focused on AI security to meet growing industry demand. More than 7,000 volunteers are helping design the exam and define required skills. The first beta tests are planned for early 2027, with wider release in April. The certification aims to reach the same recognition as the long-standing CISSP. Experts also worry about cyber warfare, AI risks, and future quantum computing threats.
New rules keep AI training stable across devices
A new study presents two methods called CROG and CLIG to improve federated learning. Federated learning lets devices train a shared AI model without sharing raw data. Problems happen when client data differs widely, causing the model to drift. The new methods weight each device's update based on how far it strays from a reference point. The approach requires no extra data, computation, or communication.
Cadence Maestro uses AI agents for chip design
Cadence Design Systems launched Maestro, a set of AI agents built for chip design. Maestro uses specialized models, reinforcement learning, and EDA tools to improve engineering workflows. It works with ChipStack, InnoStack, ViraStack, and AuraStack AI Super Agents. Cadence reports up to 30% better results and five times more token efficiency. NVIDIA is collaborating on agent security using Nemotron model technology.
AI bots keep calling local businesses and owners are unsure what to think
AI agents are calling bars, bike shops, and restaurants in San Francisco to collect data or book appointments. A test of 848 calls found that about 1 in 3 people kept talking after learning they were speaking with AI. Some business owners say the calls are annoying and get stuck during payments. Others are still figuring out if this is a problem or a new part of doing business.
Why AI tools stall when security teams move from tests to real use
A Cyberfuturists report finds that AI in security operations centers faces a gap between pilot projects and real production use. The study highlights five barriers: trust, governance, integration, cost, and verification. Security teams often let AI summarize alerts but hesitate to let it take action. The report says AI needs to prove it works under real conditions before teams fully rely on it.
Psychologist to discuss AI use in mental health at Campbell lecture
Dr. Samuel Faulkner, a 2013 Campbell University graduate and licensed psychologist, will speak on Feb. 8 about AI in mental health. His talk is called “Faster Than the Evidence: What AI Is Already Doing in Mental Health.” He focuses on how AI tools handle crisis care, screen for risk, and raise ethical questions. He says AI is moving faster than research can verify whether it works safely.
First fully AI-generated movie flops at China box office
The AI-made sci-fi film “Sanxingdui: Future Past” earned only about $198,000 in China despite over 10,000 opening-day screenings. It was the first long AI movie to get a wide release with reported box office numbers. The film cost about $1.5 million to make. Audience members said the quality still falls short of human-made movies, though the production company called it an important test.
TÜV Rheinland pushes human factors research early in product design
TÜV Rheinland spoke at the HFE & AI 2026 conference in Beijing on Oct. 17, 2026. Expert Mia Xie said human factors research should start at the beginning of product development, not just at the end. She explained that better technical specs do not always mean a better user experience. The company wants teams to use independent research to guide design, testing, and smart product decisions.
University of Arkansas professors discuss AI benefits and risks in Fayetteville
Professors at the University of Arkansas held an event in Fayetteville to talk about artificial intelligence. They explored the promising benefits AI can offer. The professors also addressed the potential dangers and challenges that come with the technology.
Sources
- As AI in politics becomes more common, here's how voters can stay vigilant
- Some voters turn to AI in making election decisions
- ISC2 Launches New AI Security Certification as Cyber Skills Demands Grow
- New Federated Learning Aggregation Rules Keep AI Training Stable When Client Data Diverge
- Cadence Maestro puts domain-trained AI agents to work for chip design
- Press 1 if you’re a robot: AI agents keep calling local businesses
- Reasoning Meets Reliability: Why AI Stalls in the Modern SOC
- Alumnus to speak on 'AI in mental health' for February Paris Lecture
- First Fully AI-Generated Movie Flops at China Box Office
- HFE & AI 2026: TÜV Rheinland Advocates Human Factors Research Across Full Product Development Cycle
- Professors at University of Arkansas weigh AI’s promise, potential dangers during Fayetteville event
Comments
Please log in to post a comment.