Cisco has rolled out new AI security capabilities that put Anthropic's Claude Mythos 5 to work inside its AI Defense platform. The system uses adaptive red teaming to simulate real-world attacks on AI agents, searching for multi-step attack paths and subtle vulnerabilities across complex setups. Standard checks cover common threats like prompt injection and data exposure, while adaptive mode lets users set custom tests in plain language and receive tailored attack reports. All actions are logged for auditing, and Cisco says the design keeps humans in control of powerful AI tools.
Customers do not get direct access to Claude Mythos; they interact with it through the AI Defense interface. Cisco also launched AI IQ, which checks live threat data against network infrastructure to speed up fixes, and tools that help organizations track AI spending and connect costs to real returns. Together, these products reflect a push to make AI security more practical and measurable.
In a separate development, the University of Nebraska-Lincoln will lead cybersecurity for FabAID, a national AI research project run by the Morgridge Institute for Research at the University of Wisconsin-Madison. The project received $24.5 million over five years from the National Science Foundation to connect AI tools and data to universities across the United States. The Holland Computing Center will build security practices and provide the computing infrastructure needed to keep the system running.
Meanwhile, the AI cloud market saw mixed results recently. Cloudflare stock jumped 11% after reporting strong Q3 2026 results, with revenue growing 39.2% year over year to $782.51 million. CEO Matthew Prince called it the company's best quarter ever. Nebius rose 9% ahead of its own report, and CoreWeave gained 6% after a deal with Wayve to use NVIDIA Blackwell infrastructure. These AI cloud companies gained ground even as NVIDIA shares dropped, though concerns about Cloudflare's gross margin and convertible debt remain.
On the policy front, Pennsylvania signed the Artificial Intelligence Deceptive Advertising Act on October 22, 2026. The law stops ads from using AI-generated content that misleads consumers and requires a clear disclosure when content is made or changed by AI. Violations can lead to fines up to $3,000. However, the law leaves open questions about enforcement, including who counts as the advertiser and whether private lawsuits are allowed.
Economic concerns are also shaping the AI conversation. An economist predicts AI investment will slow next year and fall by about 10% in 2028, even as hyperscalers are expected to spend $800 billion this year and possibly $1 trillion or more in 2027. She says AI revenue would need to grow tenfold by 2030 to justify that spending, and the S&P 500 could be 12% lower by the end of 2027. Still, she expects AI to keep improving productivity at about 1% per year.
In practical applications, AI-powered air traffic control software called SMART is reducing flight delays at DFW International Airport. The system uses 200 data streams, including weather and flight paths, to find faster routes. It has rerouted nearly 180 flights, improved flight times by over 10 minutes, and saved about 80 gallons of fuel on average per rerouted flight. The tool has been expanded to several U.S. airports.
Buffett and Munger continue to offer lessons for investors navigating the AI boom. They passed on internet stocks in the 1990s and focused on stable, profitable companies instead, a quality investing strategy that protected them during the dot-com crash. Their approach suggests that patience and a focus on profitability matter when new technologies emerge.
Key Takeaways
- Cisco integrated Anthropic's Claude Mythos 5 into its AI Defense platform for adaptive red teaming of AI agents, with all actions logged for auditing.
- FabAID, a national AI research project, received $24.5 million over five years from the National Science Foundation, led by the University of Nebraska-Lincoln.
- Cloudflare reported Q3 2026 revenue of $782.51 million, up 39.2% year over year, while AI cloud stocks rose even as NVIDIA shares fell.
- Pennsylvania's new Artificial Intelligence Deceptive Advertising Act requires clear disclosure of AI-generated ad content, with fines up to $3,000 for violations.
- Hyperscalers are expected to spend $800 billion on AI this year, but an economist warns investment could fall 10% by 2028.
- The SMART AI air traffic software has rerouted nearly 180 flights at DFW International Airport, improving flight times by over 10 minutes.
- Cisco's AI IQ tool checks live threat data against network infrastructure to speed up security fixes.
- The Pennsylvania AI ad law leaves enforcement questions unresolved, including who counts as the advertiser and whether private lawsuits are permitted.
- Warren Buffett's quality investing approach, which avoided internet stocks in the 1990s, offers guidance for investors navigating the current AI boom.
- CoreWeave gained 6% after a deal with Wayve to use NVIDIA Blackwell infrastructure for AI computing.
Cisco launches AI red teaming and observability tools
Cisco introduced new AI security tools including AI Defense, Splunk, and Cisco IQ. The system uses adaptive red teaming with Anthropic's Claude Mythos to simulate real attacks on AI agents. Organizations can now track AI spending and connect costs to real returns. Cisco IQ checks live threat data against network infrastructure to speed up fixes.
Cisco adds Claude Mythos to AI Defense
Cisco integrated Anthropic's Claude Mythos 5 into its AI Defense platform for stronger red teaming. Customers do not get direct access to Mythos but use it through the AI Defense interface. The system can now find multi-step attack paths and subtle vulnerabilities in complex setups. All actions are logged for auditing. Cisco says this approach gives humans control over powerful AI tools.
Cisco pairs Claude Mythos with AI Defense red teaming
Cisco announced that AI Defense now uses Anthropic's Claude Mythos for adaptive AI red teaming. The system tests AI agents against risks specific to each organization. Standard red teaming checks common problems like prompt injection and data exposure. Adaptive red teaming lets users set custom tests in plain language and gets tailored attack reports. Claude Mythos 5 strengthens the reasoning behind those custom tests.
Pennsylvania bans deceptive AI ads without disclosure
Pennsylvania signed the Artificial Intelligence Deceptive Advertising Act on October 22, 2026. The law stops ads from using AI-generated content that misleads consumers about products or services. Advertisers must add a clear disclosure when content is made or changed by AI. The disclosure must appear in the same medium and be easy to notice. Violations can lead to fines up to $3,000.
Pennsylvania AI ad law leaves questions on enforcement
A new Pennsylvania law requires ads with AI-generated content to include a clear disclosure. The law does not clearly define who counts as the advertiser. Experts debate whether the business, ad agency, or tech company is responsible. The Pennsylvania Attorney General can bring cases, but no private lawsuits are allowed. Enforcement may take time as officials learn how to apply the law.
Dealers turn to AI and buy centers for used car inventory
Dealers are finding it harder to get used cars at auctions because fees are rising and competition is growing. Many are turning to trade-ins and their service drives instead. One dealer sells eight used cars for every new one and built his own buy center. Another dealer uses an AI tool called AutoAcquire AI to contact thousands of customers each month. The goal is to find quality inventory without relying so much on auctions.
University of Nebraska-Lincoln to lead cybersecurity in national AI project
The University of Nebraska-Lincoln will lead cybersecurity for a national AI research project called FabAID. The project is run by the Morgridge Institute for Research at the University of Wisconsin-Madison. It aims to connect AI tools and data to universities across the United States. FabAID received $24.5 million over five years from the National Science Foundation. The Holland Computing Center will build security practices and provide the computing infrastructure needed to keep the system running.
Cloudflare and AI cloud stocks rise as NVIDIA shares fall
Cloudflare stock jumped 11% after reporting strong Q3 2026 results. Revenue grew 39.2% year over year to $782.51 million. CEO Matthew Prince called it the company's best quarter ever. Nebius rose 9% ahead of its Q3 report, and CoreWeave gained 6% after a deal with Wayve to use NVIDIA Blackwell infrastructure. These AI cloud companies gained value even as NVIDIA shares dropped. The report also noted concerns about Cloudflare's gross margin and convertible debt.
AI air traffic software helps cut flight delays at Dallas airport
New AI-powered air traffic control software called SMART is reducing flight delays at DFW International Airport. The system uses 200 data streams including weather and flight paths to find faster routes. In one case, it helped a cargo plane arrive 12 minutes early. The tool has been expanded to several U.S. airports and has rerouted nearly 180 flights. Officials say it has improved flight times by over 10 minutes and saved about 80 gallons of fuel on average per rerouted flight.
Economist warns AI investment slowdown may hurt stocks and GDP
An economist predicts AI investment will slow next year and fall by about 10% in 2028. She says companies are spending heavily, with hyperscalers expected to spend $800 billion this year and possibly $1 trillion or more in 2027. She believes AI revenue would need to grow tenfold by 2030 to justify that spending. The S&P 500 could be 12% lower by the end of 2027, she added. Still, she expects AI to keep improving productivity at about 1% per year.
AI and physical AI development calls for careful technology planning
CEO Lee focused on the fortune counseling market and its growth potential. The rise of shamanistic entertainment programs has influenced this area. As artificial intelligence and physical AI develop, careful planning for technology areas is needed. The article was translated using AI.
Buffett and Munger offer lessons for investing in AI
Warren Buffett and Charlie Munger provide guidance for investors navigating the AI boom. They passed on internet stocks in the 1990s and focused on stable, profitable companies instead. Their quality investing strategy protected them during the dot-com crash. The same questions about AI profits and impact are being asked today. Their approach shows that patience and focus on profitability matter when new technologies emerge.
Sources
- Cisco Unveils Frontier AI Red Teaming and Agent Observability to Secure the Agentic Enterprise
- Cisco tames Mythos, integrates it into AI Defense
- Cisco AI Defense to Add Anthropic’s Claude Mythos for AI Red Teaming
- Pennsylvania Enacts Artificial Intelligence Deceptive Advertising Act
- Pennsylvania wants ads to disclose AI use, but what exactly does that mean?
- Dealer panel points out increasing role of AI in used vehicle acquisition
- University of Nebraska-Lincoln to lead cybersecurity in national AI project
- Cloudflare Jumps 11%, Nebius Surges 9% as AI Cloud Stocks Defy NVIDIA's Slide; CoreWeave Rallies 6% on Wayve Deal
- Feds Say AI Software Reducing Dallas Flight Delays
- Stocks will tumble much further as AI investment will slow next year then fall in 2028 and drag GDP down too, economist predicts
- "As artificial intelligence (AI) and physical AI develop, we need to think about areas that technolo..
- Buffett And Munger Would Know How To Invest In AI
Comments
Please log in to post a comment.