CrowdStrike is riding a wave of AI security demand, with the stock up 95.5% over the past year. However, analysts believe the current price of $245.70 may be 27% overvalued compared to a fair value estimate of $193. The company's growth hinges on continued interest in cloud-native security and Next-Gen SIEM products.
Open Text has teamed up with Cohere to develop AI tools tailored for governments and regulated industries. The stock sits at $23.52, below a fair value of $27.95, and is down 25.9% year to date. Strong cloud renewal rates and a growing sales pipeline are positives, though restructuring costs remain a concern.
Workday has partnered with Databricks to build a governed AI framework using Apache Iceberg and Unity Catalog. This setup addresses the mess of point-to-point connections that AI agents create across hundreds of data systems. Unity Catalog keeps governance, access control, and auditing centralized in one place rather than duplicated for each agent.
Mantic, an AI forecasting startup, raised $25 million in seed funding after beating human contestants in the Metaculus Cup. Founded in 2024 by Toby Shevlane and Ben Day, both formerly of Google DeepMind, the company tests its frontier models on historical data and has attracted interest from businesses and government agencies worldwide.
A Forbes report reveals that only 6% of companies are extracting real value from AI. Gartner projects that specialized models will grow from $1.1 billion in spending in 2025 to more than half of all model use by 2028, signaling a major shift in how enterprises deploy AI. Governance and security remain unresolved hurdles.
On the regulatory front, seven large AI operators in Hunan, China were summoned on September 18 for violating cybersecurity and generative AI service rules. Five completed rectification while two took their models offline. Meanwhile, a debate on AI regulation intensified in the U.S., with discussions centering on balancing innovation against oversight amid competition with China.
Key Takeaways
- CrowdStrike stock is up 95.5% over the past year but may be 27% overvalued at $245.70 versus a fair value of $193.
- Open Text partners with Cohere to build AI tools for governments and regulated industries, with the stock trading below fair value at $23.52.
- Workday and Databricks created a unified governed data layer using Apache Iceberg and Unity Catalog for over 10,000 organizations.
- Mantic raised $25 million in seed funding after beating humans in the Metaculus Cup forecasting competition.
- Only 6% of companies report real value from AI, per a Forbes report.
- Gartner forecasts specialized AI model spending will grow from $1.1 billion in 2025 to over half of all model use by 2028.
- Seven AI operators in Hunan, China were summoned for security violations on September 18, with two models taken offline.
- Vision-language models present a new security blind spot, as poisoned images can bypass traditional defenses without needing a password.
- A new benchmark called TranSGrid found that the best AI models solved only 55.3% of novel reasoning tasks versus 79.6% on standard tests.
- AI regulation debate intensifies in the U.S. as policymakers weigh innovation against oversight amid competition with China.
CrowdStrike rides AI security demand but may be overvalued
CrowdStrike is benefiting from strong demand for AI security tools and new partnerships. The stock has risen sharply, gaining 17.6% in seven days and 95.5% over one year. Analysts estimate a fair value of $193, suggesting the stock at around $245.70 may be 27% overvalued. Future growth depends on continued demand for cloud-native security and Next-Gen SIEM products.
Open Text partners with Cohere for secure AI tools
Open Text announced a partnership with Cohere to build AI tools for governments and regulated industries. The stock has gained 4.02% over seven days but is down 25.9% year to date. Its current price of $23.52 is below a fair value estimate of $27.95. Strong cloud renewal rates and a growing sales pipeline support future growth, though restructuring costs remain a risk.
Vision-language models create new AI security blind spot
Security teams have focused on blocking malicious text prompts but now face a gap in protecting vision-language models. These models process images like invoices, screenshots, and scanned PDFs that can carry hidden attacks. A poisoned image needs no password and can slip into many enterprise workflows. The growing use of vision-language models makes this a serious and scalable security risk.
Only 6% of companies get real value from AI
A Forbes report says only 6% of companies are seeing real value from AI. Enterprises are shifting from single models to a mix of frontier, open, and specialized models. Gartner forecasts specialized models will grow from $1.1 billion of $14.2 billion in spending in 2025 to more than half of all model use by 2028. Governance, security, and clear ownership of AI remain major unresolved challenges.
AI tests miss real reasoning challenges, study finds
A new study introduces TranSGrid to test how well AI models handle novel problem solving. Seven Transformer models were tested on 4,800 problems that combine deductive, inductive, and abductive reasoning. The best model solved 79.6% of standard tests but only 55.3% of TranSGrid tasks. The findings show that current tests oversimplify what true systematic generalization requires.
Current launches AI tool for MSP sales and project management
Kevin Damghani, CEO of ITPartners+, launched Current, a new AI-focused company for MSPs. Current works as an overlay on existing PSA platforms like Halo, ConnectWise and Autotask. It uses AI to analyze customer data, find sales opportunities and automate business reviews. About 20 MSPs have signed up since the soft launch on September 1. The tool aims to give MSP employees a simpler, more modern interface.
Workday builds governed AI using Unity Catalog and Iceberg
Workday partnered with Databricks to create a universal data layer using Apache Iceberg and Unity Catalog. This approach solves the problem of AI agents creating complex point-to-point data connections across 100 agents and 100 data systems. Unity Catalog provides centralized governance, access control and auditing. The design keeps data governed in one place rather than replicating it for each agent. Workday serves over 10,000 organizations worldwide.
Seven AI operators in Hunan summoned for security violations
Seven large AI model operators in Hunan, China were summoned by the local cyberspace regulator on September 18, 2026. They were cited for information-security problems with their generative AI services. The operators violated China's Cybersecurity Law and Interim Measures for the Management of Generative AI Services. Five operators completed rectification while two took their models offline. They were ordered to identify security vulnerabilities and implement fixes.
Mantic raises $25 million for AI forecasting technology
Mantic, an AI startup, raised $25 million in seed funding on September 18. The company's AI beat human contestants in the Metaculus Cup forecasting competition. Mantic specializes frontier AI models for predicting events and tests them on historical data. CEO Toby Shevlane co-founded the company in 2024 with Ben Day after working at Google DeepMind. The startup has drawn interest from companies and government agencies globally.
AI regulation debate intensifies amid competition with China
A debate over artificial intelligence regulation took place on September 18, 2026. Former White House deputy chief of staff Taylor Budowich discussed the issue after Senator Rand Paul blocked Senator John Kennedy's proposed AI bill on The Evening Edit. The conversation highlighted growing stakes in the race with China over AI regulation. The discussion centered on balancing innovation with oversight.
Sources
- CrowdStrike Holdings (CRWD) Rides AI Security Demand, Is It 27% Overvalued?
- Is Open Text (OTEX) Below Fair Value After Its Cohere AI Partnership?
- What Enters Through an Image: The Blind Spot in AI Security Nobody is Detecting
- Only 6% of companies get value from AI, Forbes reports
- What Do Current Systematic Generalization Tasks Miss? A Reasoning-Centered Analysis
- MSP CEO Launches Current, A New Company Aimed At Bringing AI-First Automation To MSP Sales, Projects
- How Workday Builds Governed Agentic AI with Unity Catalog
- Seven AI-model operators in Hunan, China, summoned over security lapses | MLex | Specialist news and analysis on legal risk and regulation
- AI startup Mantic raises $25 million for superhuman forecasting
- AI regulation battle raises stakes in race with China
Comments
Please log in to post a comment.