California is rapidly deploying artificial intelligence to manage traffic congestion. Monterey County launched a pilot project costing about $1.2 million to install AI-controlled signals on a 9-mile stretch of Highway 68 near WeatherTech Raceway Laguna Seca. Simultaneously, the California Department of Transportation announced plans to use AI on a 30-mile section of Interstate 10 in Los Angeles, aiming to save hundreds of millions of dollars in construction costs by the end of the year.
While transportation agencies adopt these tools, the security implications of AI are becoming a major concern. Palo Alto Networks discovered that AI tools can identify security flaws faster than human teams, scanning code in three weeks where humans might take a year. This speed advantage is now a double-edged sword, as attackers could exploit these same tools to find vulnerabilities in critical infrastructure.
Anthropic released a powerful model called Mythos designed to find software bugs, but its high cost and potential for misuse by hackers have raised alarms among experts. Wallarm is responding by launching new assessment tools to help companies protect their APIs and AI systems from evolving threats like tool poisoning. Meanwhile, Qualcomm is preparing for a fierce competition in the AI hardware market, developing chips for phones, wearables, and personal AI companions to ensure silicon remains central to these platforms.
Research continues to evolve how AI functions and is applied. Scientists introduced Delta-Code Generation to reduce neural network search costs by 75% to 85%, making advanced research more accessible. A new brain study also revealed that decision-making signals appear earlier in the somatosensory cortex than previously thought, offering a blueprint for more efficient, low-power AI designs. However, practical applications still face hurdles; tests showed that AI trip planners like Google Gemini and ChatGPT often miss local details compared to human experts.
Key Takeaways
['Monterey County launched a $1.2 million AI traffic control pilot on Highway 68 to reduce congestion near WeatherTech Raceway Laguna Seca.', 'The California Department of Transportation will deploy AI on a 30-mile stretch of Interstate 10, potentially saving hundreds of millions in construction costs.', 'Palo Alto Networks found AI tools can scan for security flaws in three weeks, a task that typically takes human teams a full year.', 'Anthropic released the Mythos AI model, which excels at finding vulnerabilities but poses significant hacking risks due to its power and cost.', 'Wallarm is launching new tools to assess API and AI security maturity as traditional methods struggle to keep pace with fast-changing AI environments.', 'Qualcomm is developing chips for next-generation AI devices, including wearables and personal companions, to secure its position in the hardware market.', 'Researchers introduced Delta-Code Generation, a method that reduces neural network search costs by 75% to 85% while improving accuracy.', 'A new brain study shows decision signals appear in the primary somatosensory cortex, suggesting a new blueprint for more efficient AI architecture.', 'Tests revealed that AI trip planners like Google Gemini and ChatGPT often provide outdated information and miss key local spots compared to human planners.', 'Experts warn that while AI accelerates bug finding, the resulting gap between finding and fixing vulnerabilities could leave systems exposed for longer periods.']California Highway 68 Gets AI Traffic Control
Monterey County launched a new AI system to manage traffic on Highway 68. Nine signals between San Benancio Road and Josselyn Canyon Road now use sensors to adjust timing automatically. This 9-mile stretch often gets very crowded during summer events at WeatherTech Raceway Laguna Seca. The pilot project costs about $1.2 million, which is much cheaper than building roundabouts. Officials hope the technology will reduce congestion and improve travel times for drivers.
Los Angeles Highway 10 to Use AI for Traffic
The California Department of Transportation announced it will use artificial intelligence to manage traffic on Interstate 10 in Los Angeles County. The system covers a 30-mile stretch and uses real-time data to adjust signals and optimize flow. Officials say this approach could reduce congestion and save hundreds of millions of dollars in construction costs. The project is expected to be finished by the end of the year. Other states like Texas and Florida have already started using similar AI systems.
AI Cyber Tools Find More Bugs Than Human Teams
Palo Alto Networks found that AI tools can scan code faster and find more security flaws than human testers. Their tests showed three weeks of AI analysis covered more ground than a full year of manual testing. The AI models can link small vulnerabilities together to create big security risks in as little as 25 minutes. Experts warn that attackers will soon have access to these powerful tools, forcing defenders to work at machine speed. Companies must upgrade their security systems now to stay ahead of these new threats.
Wallarm Launches New API and AI Security Tools
Wallarm is focusing on protecting APIs and AI systems with new assessment tools and field events. The company says traditional security tests are not enough for fast-changing AI environments. They are hosting global sessions in May to help companies understand new AI-related risks like tool poisoning. Wallarm also offers a free assessment to help businesses check their API security maturity. Representatives attended six cybersecurity events in April to discuss these growing challenges with experts.
New AI Method Cuts Neural Network Search Costs
Researchers introduced a new method called Delta-Code Generation to lower the cost of finding optimal neural network architectures. Instead of building models from scratch, this technique uses AI to make small, efficient changes to existing designs. The new approach reduces code output by 75% to 85% while improving accuracy on various datasets. Tests showed it outperforms older methods that generate full models from the beginning. This makes advanced AI research more accessible and less expensive for scientists.
AI Translation May Harm Cultural Understanding
An article argues that perfect AI translation could destroy the deep cultural connections people make through language learning. The author shares a story about interpreting at a religious council where acting as a priest helped convey the message better than words alone. While AI tools like DeepL can translate voice instantly, they might remove the curiosity and passion that come from studying a new culture. The writer worries that languages will become just codes to decode rather than windows into other worlds. True understanding of another culture requires love and effort that machines cannot provide.
Qualcomm Prepares for AI Device Competition
Qualcomm CEO Cristiano Amon says the company is building chips for the next generation of AI devices. These include phones, wearables, glasses, and personal AI companions that go beyond simple chatbots. Qualcomm is working with major AI players on secretive hardware projects to stay competitive. The goal is to become the essential silicon layer under these new AI platforms. This shift moves the AI race from cloud computers into the objects people carry and wear every day.
Anthropic Mythos AI Poses New Hacking Risks
Anthropic released a powerful new AI model called Mythos that is excellent at finding software vulnerabilities. While defenders can use it to patch systems, attackers might use it to hack critical infrastructure and steal data. The model is very expensive to run, which is why Anthropic has not shared it publicly. Experts warn that finding and exploiting bugs is becoming easier than fixing them. This creates a dangerous short-term future where systems may stay vulnerable for longer periods.
US and China Take Different Paths in AI Race
An opinion piece explains that the United States and China are racing toward different AI goals. America focuses on creating the most powerful superintelligent technology possible with minimal regulation. China, however, aims to build practical AI systems that are tightly controlled and embedded in daily life. Chinese technology is already visible in food delivery, facial recognition, and self-driving taxis. The author suggests China’s practical approach might give them a lead in the geopolitical contest between the two nations.
Brain Study Reveals New AI Design Blueprint
Scientists found that the human brain makes decisions much earlier than previously thought, challenging how current AI is built. A new study shows decision signals appear in the primary somatosensory cortex, not just in the frontal area. This means the brain uses complex feedback loops instead of a simple one-way flow of information. Researchers led by Professor Ylasov hope to copy this efficient design to create AI that uses less power. The findings were published in the Proceedings of the National Academy of Science.
AI Trip Planners Miss Details Compared to Humans
A travel expert tested AI tools like Google Gemini and ChatGPT to plan a vacation in New York’s Finger Lakes. While the AI created a basic itinerary, it included outdated restaurant information and missed key local spots. The tools provided surface-level advice that lacked the depth of a human planner. When the expert asked the AI to incorporate specific recommendations from Travel + Leisure, the results improved significantly. This suggests AI can help with planning but still needs human guidance for the best experience.
Sources
- Heavily congested California highway will now be controlled by AI
- Heavily congested California highway will now be controlled by AI
- 3 Weeks Of AI-Assisted Cybersecurity Analysis Now Providing Broader Coverage Than Full-Year Of Manual Penetration Testing, Says Palo Alto Networks
- Wallarm Sharpens API and AI Security Focus With New Assessment and Expanded Field Engagement
- LLMs Slash Neural Architecture Search Costs
- AI will make language barriers disappear – and diminish our understanding of other cultures
- Qualcomm is positioning itself for the next AI device war
- How dangerous is Anthropic’s Mythos AI?
- Opinion | America’s A.I. Is Futuristic. China Is Just Making It Work.
- Discovery Redefines the Architecture of Thought
- Can AI Really Plan Your Vacation as Well as a Person?
Comments
Please log in to post a comment.