Radware has launched MCP Server Protection, a new security tool designed to safeguard AI agents as they connect to enterprise systems. The solution provides visibility into Model Context Protocol infrastructure and defends against threats like authentication abuse, denial-of-service attacks and protocol flooding. It integrates with Radware's Agentic AI Protection and LLM Firewall, and supports cloud, on-premises and hybrid environments including Kubernetes and hardware data centers.
Applied Optoelectronics saw its stock drop 9 percent despite reporting record Q2 2026 revenue of $191.92 million, up 86.4 percent year over year. The company earned $0.06 per share but still posted a GAAP net loss. CEO Thompson Lin forecast Q3 2026 revenue between $255 million and $290 million, saying demand outpaces capacity through mid-2027. The stock decline reflects broader concerns about AI funding, with Coherent falling 7 percent and Lumentum dropping 7 percent as well.
European cities are turning to NVIDIA accelerated computing to build AI agents and digital twins for urban planning and public services. Software partners including ABA Technology, Akila and Honeywell Technologies are deploying solutions across Belgium, Finland, France, Italy and the U.K. In France, AI agents deployed in over 500 public buildings have cut document search time from weeks to minutes, while an AI factory in Milan offers real-time translation and citizen chatbots.
Other developments round out the AI space. Benzinga and WNSTN launched a white-label AI assistant for brokerage apps, with a case study of over 88,000 trades showing active users made 2.3 times as many trades per day. Yum! Brands reported that Taco Bell's drive-thru AI bot has improved employee retention by letting workers focus on hospitality. McDonald's faces a class action lawsuit over alleged AI-based menu pricing, though the company denies using AI for that purpose. Researchers also unveiled PDRA, an attack method that tricks large language models into harmful content through multi-step reasoning.
Key Takeaways
- <ul><li>Radware launched MCP Server Protection to secure AI agent connections against authentication abuse, DoS attacks and protocol flooding across cloud, on-premises and hybrid environments.</li><li>Applied Optoelectronics reported record Q2 2026 revenue of $191.92 million but its stock fell 9% amid AI funding concerns, with CEO Thompson Lin guiding Q3 revenue to $255 million to $290 million.</li><li>European cities are using NVIDIA accelerated computing for AI agents and digital twins, with deployments in France, Italy, the U.K., Belgium, Finland and other regions.</li><li>Benzinga and WNSTN launched a white-label AI assistant for brokers; active users in a study of 88,000+ trades made 2.3 times more trades per day than non-users.</li><li>Yum! Brands says Taco Bell's drive-thru AI bot has boosted employee retention by simplifying orders so staff can focus on hospitality.</li><li>McDonald's is facing a class action lawsuit alleging inconsistent AI-based menu pricing, with one Big Mac costing 21% more at a location two miles away.</li><li>Researchers developed PDRA, an attack method that exploits multi-step reasoning in large language models to suppress safety responses and produce harmful content.</li><li>Nearly half of software companies now use MCP in production, according to Stacklok's 2026 survey, signaling rapid adoption of the protocol.</li><li>The Shenzhen Component index fell 2.0 percent to 12,111 points due to AI hardware selling, dropping 10.4 percent in the first quarter.</li><li>AI-generated recipes are flooding the internet, with 7 out of 10 people unaware they were AI-made, prompting sites like Tastes Delicious to offer verified human recipes.</li></ul>
Radware launches MCP Server Protection for AI agent security
Radware introduced MCP Server Protection, a new tool to secure AI agents as they connect to enterprise systems. The solution helps organizations see their MCP infrastructure and protect against attacks like authentication abuse and denial-of-service. It works with Radware Agentic AI Protection and LLM Firewall. The tool supports cloud, on-premises and hybrid setups. David Aviv, Radware's CTO, said MCP is becoming a critical connection point between AI agents and businesses.
Radware unveils MCP Server Protection for agentic AI integration
Radware announced its MCP Server Protection solution to help organizations safely adopt Model Context Protocol. The tool gives visibility into MCP servers and protects them from protocol abuse, flooding and other attacks. It is part of Radware's cloud security platform. According to Stacklok's 2026 survey, nearly half of software companies use MCP in production. The solution supports public clouds, private clouds, Kubernetes and hardware data centers.
Applied Optoelectronics stock drops 9% amid AI funding concerns
Applied Optoelectronics saw its stock fall 9% as fears about AI funding hit optical networking companies. Despite record Q2 2026 revenue of $191.92 million, the company faces risk because its top 10 customers account for 99% of revenue. Coherent and Lumentum also saw stock declines. CEO Thompson Lin forecast demand outpacing capacity through mid-2027. The company still posted a GAAP net loss in the quarter.
Applied Optoelectronics plunges as optical stocks sell off on AI fears
Applied Optoelectronics stock plunged 9% while Coherent fell 7% and Lumentum dropped 7% due to AI funding worries. In Q2 2026, Applied Optoelectronics reported revenue up 86.4% year over year to $191.92 million and earned $0.06 per share. CEO Lin guided Q3 2026 revenue to $255 million to $290 million. Coherent CEO Jim Anderson said customer demand shows no signs of slowing. Investors are advised to limit exposure to optical stocks given potential volatility.
European cities use NVIDIA AI agents for smarter urban operations
Cities across Europe are using NVIDIA accelerated computing to build AI agents and digital twins for urban planning and public services. Software partners including ABA Technology, Akila and Honeywell Technologies are developing solutions in Belgium, Finland, France, Italy and the U.K. In France, agents in over 500 public buildings cut document search time from weeks to minutes. In Italy, an AI factory in Milan offers real-time translation and citizen chatbots. Projects also span the U.K., Sweden and Dubai.
Shashua says Israel needs AI infrastructure investment
Amnon Shashua spoke at the Journey conference about Israel's AI future. He said innovation now requires more than just brainwork. Israeli entrepreneurs are moving to the US to follow the money. Israel must invest in AI infrastructure to stay competitive.
Shenzhen Component index drops 243 points amid AI hardware sell-off
The Shenzhen Component fell 2.0 percent to 12,111 points due to AI hardware selling. A Commerce Ministry official blamed lower demand and higher production costs. The index dropped 10.4 percent in the first quarter. The government is offering subsidies and promoting AI hardware use in healthcare and finance.
AI recipes flood internet as developers push back
AI-generated recipes are taking over the internet, confusing home cooks. A survey found 7 out of 10 people did not know recipes were AI-made. Chef Jenn Allen created Tastes Delicious, a site with 4,000 verified human recipes. Developers must vouch for their recipes and photos being AI-free. Pinterest and Google are working on better AI controls for recipes.
McDonald's faces lawsuit over AI-menu pricing
McDonald's is facing a class action lawsuit over AI-based menu pricing. A customer in Chicago filed the suit claiming inconsistent prices across locations. One example showed a Big Mac costing 21 percent more at a location two miles away. McDonald's denies using AI for menu pricing and says franchisees decide prices.
Yum! Brands says Taco Bell AI drive-thru boosts employee retention
Yum! Brands says its Taco Bell drive-thru AI bot has improved employee retention. Erika Burkhardt spoke at the Fortune Most Powerful Women Summit in Santa Barbara. She said AI makes ordering easier so workers can focus on hospitality. The fast-food industry continues exploring AI tools to improve workflow efficiency.
Benzinga and WNSTN Launch AI Assistant for Brokers
Benzinga and WNSTN launched a white-label AI conversational assistant for brokerage apps. The tool combines Benzinga's market data with WNSTN's conversational technology through an API. A case study of over 88,000 trades found active users made 2.3 times as many trades per day as non-users. Every response passes through a compliance check, but brokers must still supervise the assistant's advice.
New Attack Method Exploits Reasoning in Large Language Models
Researchers developed PDRA, an attack method that tricks large language models into producing harmful content through multi-step reasoning. PDRA creates a paradoxical narrative pairing a malicious actor with a benign task to guide the model through a three-stage analysis. The attack suppresses safety responses while activating planning-oriented terms inside the model. It achieved state-of-the-art attack success rates across diverse LLMs with high efficiency.
Sources
- Radware Launches MCP Server Protection to Secure Enterprise AI Agents
- Radware Introduces Model Context Protocol (MCP) Server Protection to Help Secure the Integration Layer of Agentic AI
- Applied Optoelectronics Dives 9% as AI Funding Fears Hit Optical Stocks; Coherent Slides 7%, Lumentum Drops 7%
- Applied Optoelectronics Dives 9% as AI Funding Fears Hit Optical Stocks; Coherent Slides 7%, Lumentum Drops 7%
- European Cities Deploy NVIDIA-Powered AI Agents to Run Smarter Urban Operations
- Shashua: Israel must invest in AI infrastructure to compete
- Shenzhen Component loses 243 points as AI hardware sells off
- AI recipes have flooded the internet, and recipe developers are building ways to push back
- McDonald's faces class action lawsuit over AI-menu pricing allegations
- Yum! Brands executive pushes back against AI hate: Taco Bell drive-thru bot has helped increase employee retention by making jobs easier
- Benzinga and WNSTN Pitch Turnkey AI to Brokers With Promise of 2.3x Trading Boost
- Narrative Paradoxes in LLM Inference: Shaping Reasoning Trajectories toward Misalignment
Comments
Please log in to post a comment.